VictoriaMetrics is developed at a fast pace, so it is recommended periodically checking the [CHANGELOG](https://docs.victoriametrics.com/CHANGELOG.html) and performing [regular upgrades](#how-to-upgrade-victoriametrics).
* VictoriaMetrics consists of a single [small executable](https://medium.com/@valyala/stripping-dependency-bloat-in-victoriametrics-docker-image-983fb5912b0d) without external dependencies.
* All the configuration is done via explicit command-line flags with reasonable defaults.
* All the data is stored in a single directory pointed by `-storageDataPath` command-line flag.
* Easy and fast backups from [instant snapshots](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282) to S3 or GCS can be done with [vmbackup](https://docs.victoriametrics.com/vmbackup.html) / [vmrestore](https://docs.victoriametrics.com/vmrestore.html) tools. See [this article](https://medium.com/@valyala/speeding-up-backups-for-big-time-series-databases-533c1a927883) for more details.
* It implements PromQL-based query language - [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html), which provides improved functionality on top of PromQL.
* It provides global query view. Multiple Prometheus instances or any other data sources may ingest data into VictoriaMetrics. Later this data may be queried via a single query.
* It provides high performance and good vertical and horizontal scalability for both [data ingestion](https://medium.com/@valyala/high-cardinality-tsdb-benchmarks-victoriametrics-vs-timescaledb-vs-influxdb-13e6ee64dd6b) and [data querying](https://medium.com/@valyala/when-size-matters-benchmarking-victoriametrics-vs-timescale-and-influxdb-6035811952d4). It [outperforms InfluxDB and TimescaleDB by up to 20x](https://medium.com/@valyala/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae).
* It [uses 10x less RAM than InfluxDB](https://medium.com/@valyala/insert-benchmarks-with-inch-influxdb-vs-victoriametrics-e31a41ae2893) and [up to 7x less RAM than Prometheus, Thanos or Cortex](https://valyala.medium.com/prometheus-vs-victoriametrics-benchmark-on-node-exporter-metrics-4ca29c75590f) when dealing with millions of unique time series (aka [high cardinality](https://docs.victoriametrics.com/FAQ.html#what-is-high-cardinality)).
* It is optimized for time series with [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate).
* It provides high data compression, so [up to 70x more data points](https://medium.com/@valyala/when-size-matters-benchmarking-victoriametrics-vs-timescale-and-influxdb-6035811952d4) may be crammed into limited storage comparing to TimescaleDB and [up to 7x less storage space is required compared to Prometheus, Thanos or Cortex](https://valyala.medium.com/prometheus-vs-victoriametrics-benchmark-on-node-exporter-metrics-4ca29c75590f).
* It is optimized for storage with high-latency IO and low IOPS (HDD and network storage in AWS, Google Cloud, Microsoft Azure, etc). See [disk IO graphs from these benchmarks](https://medium.com/@valyala/high-cardinality-tsdb-benchmarks-victoriametrics-vs-timescaledb-vs-influxdb-13e6ee64dd6b).
* A single-node VictoriaMetrics may substitute moderately sized clusters built with competing solutions such as Thanos, M3DB, Cortex, InfluxDB or TimescaleDB. See [vertical scalability benchmarks](https://medium.com/@valyala/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae), [comparing Thanos to VictoriaMetrics cluster](https://medium.com/@valyala/comparing-thanos-to-victoriametrics-cluster-b193bea1683) and [Remote Write Storage Wars](https://promcon.io/2019-munich/talks/remote-write-storage-wars/) talk from [PromCon 2019](https://promcon.io/2019-munich/talks/remote-write-storage-wars/).
* It protects the storage from data corruption on unclean shutdown (i.e. OOM, hardware reset or `kill -9`) thanks to [the storage architecture](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282).
* It supports metrics' scraping, ingestion and [backfilling](#backfilling) via the following protocols:
* [Metrics scraping from Prometheus exporters](#how-to-scrape-prometheus-exporters-such-as-node-exporter).
* It can deal with [high cardinality issues](https://docs.victoriametrics.com/FAQ.html#what-is-high-cardinality) and [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate) issues via [series limiter](#cardinality-limiter).
* It ideally works with big amounts of time series data from APM, Kubernetes, IoT sensors, connected cars, industrial telemetry, financial data and various [Enterprise workloads](https://victoriametrics.com/products/enterprise/).
* [German Research Center for Artificial Intelligence](https://docs.victoriametrics.com/CaseStudies.html#german-research-center-for-artificial-intelligence)
See also [articles and slides about VictoriaMetrics from our users](https://docs.victoriametrics.com/Articles.html#third-party-articles-and-slides-about-victoriametrics)
Just download [VictoriaMetrics executable](https://github.com/VictoriaMetrics/VictoriaMetrics/releases) or [Docker image](https://hub.docker.com/r/victoriametrics/victoria-metrics/) and start it with the desired command-line flags.
*`-retentionPeriod` - retention for stored data. Older data is automatically deleted. Default retention is 1 month. See [the Retention section](#retention) for more details.
Other flags have good enough default values, so set them only if you really need this. Pass `-help` to see [all the available flags with description and default values](#list-of-command-line-flags).
The following docs may be useful during initial VictoriaMetrics setup:
* [How to set up scraping of Prometheus-compatible targets](https://docs.victoriametrics.com/#how-to-scrape-prometheus-exporters-such-as-node-exporter)
* [How to ingest data to VictoriaMetrics](#how-to-import-time-series-data)
* [How to set up Prometheus to write data to VictoriaMetrics](https://docs.victoriametrics.com/#prometheus-setup)
* [How to query VictoriaMetrics via Grafana](#grafana-setup)
* [How to query VictoriaMetrics via Graphite API](#graphite-api-usage)
VictoriaMetrics is developed at a fast pace, so it is recommended periodically checking the [CHANGELOG](https://docs.victoriametrics.com/CHANGELOG.html) and performing [regular upgrades](#how-to-upgrade-victoriametrics).
* Each `.` char in flag name must be substituted with `_` (for example `-insert.maxQueueDuration <duration>` will translate to `insert_maxQueueDuration=<duration>`).
* For repeating flags an alternative syntax can be used by joining the different values into one using `,` char as separator (for example `-storageNode <nodeA> -storageNode <nodeB>` will translate to `storageNode=<nodeA>,<nodeB>`).
* Environment var prefix can be set via `-envflag.prefix` flag. For instance, if `-envflag.prefix=VM_`, then env vars must be prepended with `VM_`.
Add the following lines to Prometheus config file (it is usually located at `/etc/prometheus/prometheus.yml`) in order to send data to VictoriaMetrics:
Then build graphs and dashboards for the created datasource using [PromQL](https://prometheus.io/docs/prometheus/latest/querying/basics/) or [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html).
VictoriaMetrics is developed at a fast pace, so it is recommended periodically checking [the CHANGELOG page](https://docs.victoriametrics.com/CHANGELOG.html) and performing regular upgrades.
It is safe upgrading VictoriaMetrics to new versions unless [release notes](https://github.com/VictoriaMetrics/VictoriaMetrics/releases) say otherwise. It is safe skipping multiple versions during the upgrade unless [release notes](https://github.com/VictoriaMetrics/VictoriaMetrics/releases) say otherwise. It is recommended performing regular upgrades to the latest version, since it may contain important bug fixes, performance optimizations or new features.
Prometheus doesn't drop data during VictoriaMetrics restart. See [this article](https://grafana.com/blog/2019/03/25/whats-new-in-prometheus-2.8-wal-based-remote-write/) for details. The same applies also to [vmagent](https://docs.victoriametrics.com/vmagent.html).
* Drag the graph to the left / right in order to move the displayed time range into the past / future.
* Hold `Ctrl` (or `Cmd` on MacOS) and scroll up / down in order to zoom in / out the graph.
Query history can be navigated by holding `Ctrl` (or `Cmd` on MacOS) and pressing `up` or `down` arrows on the keyboard while the cursor is located in the query input field.
Multi-line queries can be entered by pressing `Shift-Enter` in query input field.
When querying the [backfilled data](https://docs.victoriametrics.com/#backfilling) or during [query troubleshooting](https://docs.victoriametrics.com/Troubleshooting.html#unexpected-query-results), it may be useful disabling response cache by clicking `Disable cache` checkbox.
VMUI automatically adjusts the interval between datapoints on the graph depending on the horizontal resolution and on the selected time range. The step value can be customized by clickhing `Override step value` checkbox.
VMUI allows investigating correlations between two queries on the same graph. Just click `+` button, enter the second query in the newly appeared input field and press `Ctrl+Enter`. Results for both queries should be displayed simultaneously on the same graph. Every query has its own vertical scale, which is displayed on the left and the right side of the graph. Lines for the second query are dashed.
See the [example VMUI at VictoriaMetrics playground](https://play.victoriametrics.com/select/accounting/1/6a716b0f-38bc-4856-90ce-448fd713e3fe/prometheus/graph/?g0.expr=100%20*%20sum(rate(process_cpu_seconds_total))%20by%20(job)&g0.range_input=1d).
See [cardinality explorer playground](https://play.victoriametrics.com/select/accounting/1/6a716b0f-38bc-4856-90ce-448fd713e3fe/prometheus/graph/#/cardinality).
Prometheus doesn't drop data during VictoriaMetrics restart. See [this article](https://grafana.com/blog/2019/03/25/whats-new-in-prometheus-2.8-wal-based-remote-write/) for details. The same applies alos to [vmagent](https://docs.victoriametrics.com/vmagent.html).
VictoriaMetrics can be used as drop-in replacement for Prometheus for scraping targets configured in `prometheus.yml` config file according to [the specification](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#configuration-file). Just set `-promscrape.config` command-line flag to the path to `prometheus.yml` config - and VictoriaMetrics should start scraping the configured targets. Currently the following [scrape_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config) types are supported:
The file pointed by `-promscrape.config` may contain `%{ENV_VAR}` placeholders, which are substituted by the corresponding `ENV_VAR` environment variable values.
VictoriaMetrics also supports [importing data in Prometheus exposition format](#how-to-import-data-in-prometheus-exposition-format).
See also [vmagent](https://docs.victoriametrics.com/vmagent.html), which can be used as drop-in replacement for Prometheus.
VictoriaMetrics accepts data from [DataDog agent](https://docs.datadoghq.com/agent/) or [DogStatsD]() via ["submit metrics" API](https://docs.datadoghq.com/api/latest/metrics/#submit-metrics) at `/datadog/api/v1/series` path.
Run DataDog agent with `DD_DD_URL=http://victoriametrics-host:8428/datadog` environment variable in order to write data to VictoriaMetrics at `victoriametrics-host` host. Another option is to set `dd_url` param at [DataDog agent configuration file](https://docs.datadoghq.com/agent/guide/agent-configuration-files/) to `http://victoriametrics-host:8428/datadog`.
* Field names are mapped to time series names prefixed with `{measurement}{separator}` value, where `{separator}` equals to `_` by default. It can be changed with `-influxMeasurementFieldSeparator` command-line flag. See also `-influxSkipSingleField` command-line flag. If `{measurement}` is empty or if `-influxSkipMeasurement` command-line flag is set, then time series names correspond to field names.
Note that InfluxDB line protocol expects [timestamps in *nanoseconds* by default](https://docs.influxdata.com/influxdb/v1.7/write_protocols/line_protocol_tutorial/#timestamp),
[Graphite relabeling](https://docs.victoriametrics.com/vmagent.html#graphite-relabeling) can be used if the imported Graphite data is going to be queried via [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html).
VictoriaMetrics supports `__graphite__` pseudo-label for selecting time series with Graphite-compatible filters in [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html). For example, `{__graphite__="foo.*.bar"}` is equivalent to `{__name__=~"foo[.][^.]*[.]bar"}`, but it works faster and it is easier to use when migrating from Graphite to VictoriaMetrics. See [docs for Graphite paths and wildcards](https://graphite.readthedocs.io/en/latest/render_api.html#paths-and-wildcards). VictoriaMetrics also supports [label_graphite_group](https://docs.victoriametrics.com/MetricsQL.html#label_graphite_group) function for extracting the given groups from Graphite metric name.
The `__graphite__` pseudo-label supports e.g. alternate regexp filters such as `(value1|...|valueN)`. They are transparently converted to `{value1,...,valueN}` syntax [used in Graphite](https://graphite.readthedocs.io/en/latest/render_api.html#paths-and-wildcards). This allows using [multi-value template variables in Grafana](https://grafana.com/docs/grafana/latest/variables/formatting-multi-value-variables/) inside `__graphite__` pseudo-label. For example, Grafana expands `{__graphite__=~"foo.($bar).baz"}` into `{__graphite__=~"foo.(x|y).baz"}` if `$bar` template variable contains `x` and `y` values. In this case the query is automatically converted into `{__graphite__=~"foo.{x,y}.baz"}` before execution.
* [/api/v1/status/tsdb](https://prometheus.io/docs/prometheus/latest/querying/api/#tsdb-stats). See [these docs](#tsdb-stats) for details.
* [/api/v1/targets](https://prometheus.io/docs/prometheus/latest/querying/api/#targets) - see [these docs](#how-to-scrape-prometheus-exporters-such-as-node-exporter) for more details.
These handlers can be queried from Prometheus-compatible clients such as Grafana or curl.
All the Prometheus querying API handlers can be prepended with `/prometheus` prefix. For example, both `/prometheus/api/v1/query` and `/api/v1/query` should work.
### Prometheus querying API enhancements
VictoriaMetrics accepts optional `extra_label=<label_name>=<label_value>` query arg, which can be used for enforcing additional label filters for queries. For example,
`/api/v1/query_range?extra_label=user_id=123&extra_label=group_id=456&query=<query>` would automatically add `{user_id="123",group_id="456"}` label filters to the given `<query>`. This functionality can be used for limiting the scope of time series visible to the given tenant. It is expected that the `extra_label` query args are automatically set by auth proxy sitting in front of VictoriaMetrics. See [vmauth](https://docs.victoriametrics.com/vmauth.html) and [vmgateway](https://docs.victoriametrics.com/vmgateway.html) as examples of such proxies.
VictoriaMetrics accepts optional `extra_filters[]=series_selector` query arg, which can be used for enforcing arbitrary label filters for queries. For example,
`/api/v1/query_range?extra_filters[]={env=~"prod|staging",user="xyz"}&query=<query>` would automatically add `{env=~"prod|staging",user="xyz"}` label filters to the given `<query>`. This functionality can be used for limiting the scope of time series visible to the given tenant. It is expected that the `extra_filters[]` query args are automatically set by auth proxy sitting in front of VictoriaMetrics. See [vmauth](https://docs.victoriametrics.com/vmauth.html) and [vmgateway](https://docs.victoriametrics.com/vmgateway.html) as examples of such proxies.
VictoriaMetrics accepts relative times in `time`, `start` and `end` query args additionally to unix timestamps and [RFC3339](https://www.ietf.org/rfc/rfc3339.txt).
For example, the following query would return data for the last 30 minutes: `/api/v1/query_range?start=-30m&query=...`.
VictoriaMetrics accepts `round_digits` query arg for `/api/v1/query` and `/api/v1/query_range` handlers. It can be used for rounding response values to the given number of digits after the decimal point. For example, `/api/v1/query?query=avg_over_time(temperature[1h])&round_digits=2` would round response values to up to two digits after the decimal point.
VictoriaMetrics accepts `limit` query arg for `/api/v1/labels` and `/api/v1/label/<labelName>/values` handlers for limiting the number of returned entries. For example, the query to `/api/v1/labels?limit=5` returns a sample of up to 5 unique labels, while ignoring the rest of labels. If the provided `limit` value exceeds the corresponding `-search.maxTagKeys` / `-search.maxTagValues` command-line flag values, then limits specified in the command-line flags are used.
By default, VictoriaMetrics returns time series for the last 5 minutes from `/api/v1/series`, while the Prometheus API defaults to all time. Use `start` and `end` to select a different time range.
VictoriaMetrics accepts `limit` query arg for `/api/v1/series` handlers for limiting the number of returned entries. For example, the query to `/api/v1/series?limit=5` returns a sample of up to 5 series, while ignoring the rest. If the provided `limit` value exceeds the corresponding `-search.maxSeries` command-line flag values, then limits specified in the command-line flags are used.
*`/api/v1/series/count` - returns the total number of time series in the database. Some notes:
* the handler scans all the inverted index, so it can be slow if the database contains tens of millions of time series;
* the handler may count [deleted time series](#how-to-delete-time-series) additionally to normal time series due to internal implementation restrictions;
*`/api/v1/status/active_queries` - returns a list of currently running queries.
*`/api/v1/status/top_queries` - returns the following query lists:
* the most frequently executed queries - `topByCount`
* queries with the biggest average execution duration - `topByAvgDuration`
* queries that took the most time for execution - `topBySumDuration`
The number of returned queries can be limited via `topN` query arg. Old queries can be filtered out with `maxLifetime` query arg.
For example, request to `/api/v1/status/top_queries?topN=5&maxLifetime=30s` would return up to 5 queries per list, which were executed during the last 30 seconds.
VictoriaMetrics tracks the last `-search.queryStats.lastQueriesCount` queries with durations at least `-search.queryStats.minQueryDuration`.
VictoriaMetrics supports data ingestion in Graphite protocol - see [these docs](#how-to-send-data-from-graphite-compatible-agents-such-as-statsd) for details.
VictoriaMetrics supports the following Graphite querying APIs, which are needed for [Graphite datasource in Grafana](https://grafana.com/docs/grafana/latest/datasources/graphite/):
VictoriaMetrics accepts optional query args: `extra_label=<label_name>=<label_value>` and `extra_filters[]=series_selector` query args for all the Graphite APIs. These args can be used for limiting the scope of time series visible to the given tenant. It is expected that the `extra_label` query arg is automatically set by auth proxy sitting in front of VictoriaMetrics. See [vmauth](https://docs.victoriametrics.com/vmauth.html) and [vmgateway](https://docs.victoriametrics.com/vmgateway.html) as examples of such proxies.
VictoriaMetrics supports `__graphite__` pseudo-label for filtering time series with Graphite-compatible filters in [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html). See [these docs](#selecting-graphite-metrics).
When configuring Graphite datasource in Grafana, the `Storage-Step` http request header must be set to a step between Graphite data points stored in VictoriaMetrics. For example, `Storage-Step: 10s` would mean 10 seconds distance between Graphite datapoints stored in VictoriaMetrics.
*`label` - for selecting arbitrary label values. By default `label=__name__`, i.e. metric names are selected.
*`delimiter` - for using different delimiters in metric name hierachy. For example, `/metrics/find?delimiter=_&query=node_*` would return all the metric name prefixes
that start with `node_`. By default `delimiter=.`.
2. Run `make victoria-metrics-linux-arm` or `make victoria-metrics-linux-arm64` from the root folder of [the repository](https://github.com/VictoriaMetrics/VictoriaMetrics).
It builds `victoria-metrics-linux-arm` or `victoria-metrics-linux-arm64` binary respectively and puts it into the `bin` folder.
2. Run `make victoria-metrics-linux-arm-prod` or `make victoria-metrics-linux-arm64-prod` from the root folder of [the repository](https://github.com/VictoriaMetrics/VictoriaMetrics).
It builds `victoria-metrics-linux-arm-prod` or `victoria-metrics-linux-arm64-prod` binary respectively and puts it into the `bin` folder.
helps to spin up VictoriaMetrics, [vmagent](https://docs.victoriametrics.com/vmagent.html) and Grafana with one command.
More details may be found [here](https://github.com/VictoriaMetrics/VictoriaMetrics/tree/master/deployment/docker#folder-contains-basic-images-and-tools-for-building-and-running-victoria-metrics-in-docker).
VictoriaMetrics can create [instant snapshots](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282)
for all the data stored under `-storageDataPath` directory.
Navigate to `http://<victoriametrics-addr>:8428/snapshot/create` in order to create an instant snapshot.
The page will return the following JSON response:
```json
{"status":"ok","snapshot":"<snapshot-name>"}
```
Snapshots are created under `<-storageDataPath>/snapshots` directory, where `<-storageDataPath>`
is the command-line flag value. Snapshots can be archived to backup storage at any time
with [vmbackup](https://docs.victoriametrics.com/vmbackup.html).
The `http://<victoriametrics-addr>:8428/snapshot/list` page contains the list of available snapshots.
Navigate to `http://<victoriametrics-addr>:8428/snapshot/delete?snapshot=<snapshot-name>` in order
to delete `<snapshot-name>` snapshot.
Navigate to `http://<victoriametrics-addr>:8428/snapshot/delete_all` in order to delete all the snapshots.
Steps for restoring from a snapshot:
1. Stop VictoriaMetrics with `kill -INT`.
2. Restore snapshot contents from backup with [vmrestore](https://docs.victoriametrics.com/vmrestore.html)
to the directory pointed by `-storageDataPath`.
3. Start VictoriaMetrics.
## How to delete time series
Send a request to `http://<victoriametrics-addr>:8428/api/v1/admin/tsdb/delete_series?match[]=<timeseries_selector_for_delete>`,
where `<timeseries_selector_for_delete>` may contain any [time series selector](https://prometheus.io/docs/prometheus/latest/querying/basics/#time-series-selectors)
for metrics to delete. After that all the time series matching the given selector are deleted. Storage space for
the deleted time series isn't freed instantly - it is freed during subsequent [background merges of data files](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282).
Note that background merges may never occur for data from previous months, so storage space won't be freed for historical data.
In this case [forced merge](#forced-merge) may help freeing up storage space.
It is recommended verifying which metrics will be deleted with the call to `http://<victoria-metrics-addr>:8428/api/v1/series?match[]=<timeseries_selector_for_delete>`
VictoriaMetrics performs [data compactions in background](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282)
in order to keep good performance characteristics when accepting new data. These compactions (merges) are performed independently on per-month partitions.
This means that compactions are stopped for per-month partitions if no new data is ingested into these partitions.
Sometimes it is necessary to trigger compactions for old partitions. For instance, in order to free up disk space occupied by [deleted time series](#how-to-delete-time-series).
In this case forced compaction may be initiated on the specified per-month partition by sending request to `/internal/force_merge?partition_prefix=YYYY_MM`,
where `YYYY_MM` is per-month partition name. For example, `http://victoriametrics:8428/internal/force_merge?partition_prefix=2020_08` would initiate forced
merge for August 2020 partition. The call to `/internal/force_merge` returns immediately, while the corresponding forced merge continues running in background.
Forced merges may require additional CPU, disk IO and storage space resources. It is unnecessary to run forced merge under normal conditions,
since VictoriaMetrics automatically performs [optimal merges in background](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282)
when new data is ingested into it.
## How to export time series
VictoriaMetrics provides the following handlers for exporting data:
*`/api/v1/export` for exporing data in JSON line format. See [these docs](#how-to-export-data-in-json-line-format) for details.
*`/api/v1/export/csv` for exporting data in CSV. See [these docs](#how-to-export-csv-data) for details.
Send a request to `http://<victoriametrics-addr>:8428/api/v1/export?match[]=<timeseries_selector_for_export>`,
where `<timeseries_selector_for_export>` may contain any [time series selector](https://prometheus.io/docs/prometheus/latest/querying/basics/#time-series-selectors)
for metrics to export. Use `{__name__!=""}` selector for fetching all the time series.
The response would contain all the data for the selected time series in [JSON streaming format](https://en.wikipedia.org/wiki/JSON_streaming#Line-delimited_JSON).
Each JSON line contains samples for a single time series. An example output:
Send a request to `http://<victoriametrics-addr>:8428/api/v1/export/csv?format=<format>&match=<timeseries_selector_for_export>`,
where:
*`<format>` must contain comma-delimited label names for the exported CSV. The following special label names are supported:
*`__name__` - metric name
*`__value__` - sample value
*`__timestamp__:<ts_format>` - sample timestamp. `<ts_format>` can have the following values:
*`unix_s` - unix seconds
*`unix_ms` - unix milliseconds
*`unix_ns` - unix nanoseconds
*`rfc3339` - [RFC3339](https://www.ietf.org/rfc/rfc3339.txt) time
*`custom:<layout>` - custom layout for time that is supported by [time.Format](https://golang.org/pkg/time/#Time.Format) function from Go.
*`<timeseries_selector_for_export>` may contain any [time series selector](https://prometheus.io/docs/prometheus/latest/querying/basics/#time-series-selectors)
for metrics to export.
Optional `start` and `end` args may be added to the request in order to limit the time frame for the exported data. These args may contain either
unix timestamp in seconds or [RFC3339](https://www.ietf.org/rfc/rfc3339.txt) values.
The [deduplication](#deduplication) is applied for the data exported in CSV by default. It is possible to export raw data without de-duplication by passing `reduce_mem_usage=1` query arg to `/api/v1/export/csv`.
Send a request to `http://<victoriametrics-addr>:8428/api/v1/export/native?match[]=<timeseries_selector_for_export>`,
where `<timeseries_selector_for_export>` may contain any [time series selector](https://prometheus.io/docs/prometheus/latest/querying/basics/#time-series-selectors)
for metrics to export. Use `{__name__=~".*"}` selector for fetching all the time series.
On large databases you may experience problems with limit on the number of time series, which can be exported. In this case you need to adjust `-search.maxExportSeries` command-line flag:
The [deduplication](#deduplication) isn't applied for the data exported in native format. It is expected that the de-duplication is performed during data import.
* [Prometheus remote_write API](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#remote_write). See [these docs](#prometheus-setup) for details.
* Graphite plaintext protocol. See [these docs](#how-to-send-data-from-graphite-compatible-agents-such-as-statsd) for details.
* OpenTSDB telnet put protocol. See [these docs](#sending-data-via-telnet-put-protocol) for details.
* OpenTSDB http `/api/put` protocol. See [these docs](#sending-opentsdb-data-via-http-apiput-requests) for details.
*`/api/v1/import` for importing data obtained from [/api/v1/export](#how-to-export-data-in-json-line-format).
See [these docs](#how-to-import-data-in-json-line-format) for details.
*`/api/v1/import/native` for importing data obtained from [/api/v1/export/native](#how-to-export-data-in-native-format).
See [these docs](#how-to-import-data-in-native-format) for details.
*`/api/v1/import/csv` for importing arbitrary CSV data. See [these docs](#how-to-import-csv-data) for details.
*`/api/v1/import/prometheus` for importing data in Prometheus exposition format. See [these docs](#how-to-import-data-in-prometheus-exposition-format) for details.
VictoriaMetrics parses input JSON lines one-by-one. It loads the whole JSON line in memory, then parses it and then saves the parsed samples into persistent storage. This means that VictoriaMetrics can occupy big amounts of RAM when importing too long JSON lines. The solution is to split too long JSON lines into smaller lines. It is OK if samples for a single time series are split among multiple JSON lines.
The specification of VictoriaMetrics' native format may yet change and is not formally documented yet. So currently we do not recommend that external clients attempt to pack their own metrics in native format file.
If you have a native format file obtained via [/api/v1/export/native](#how-to-export-data-in-native-format) however this is the most efficient protocol for importing data in.
Note that it could be required to flush response cache after importing historical data. See [these docs](#backfilling) for detail.
### How to import CSV data
Arbitrary CSV data can be imported via `/api/v1/import/csv`. The CSV data is imported according to the provided `format` query arg.
The `format` query arg must contain comma-separated list of parsing rules for CSV fields. Each rule consists of three parts delimited by a colon:
```
<column_pos>:<type>:<context>
```
*`<column_pos>` is the position of the CSV column (field). Column numbering starts from 1. The order of parsing rules may be arbitrary.
*`<type>` describes the column type. Supported types are:
*`metric` - the corresponding CSV column at `<column_pos>` contains metric value, which must be integer or floating-point number.
The metric name is read from the `<context>`. CSV line must have at least a single metric field. Multiple metric fields per CSV line is OK.
*`label` - the corresponding CSV column at `<column_pos>` contains label value. The label name is read from the `<context>`.
CSV line may have arbitrary number of label fields. All these labels are attached to all the configured metrics.
*`time` - the corresponding CSV column at `<column_pos>` contains metric time. CSV line may contain either one or zero columns with time.
If CSV line has no time, then the current time is used. The time is applied to all the configured metrics.
The format of the time is configured via `<context>`. Supported time formats are:
*`unix_s` - unix timestamp in seconds.
*`unix_ms` - unix timestamp in milliseconds.
*`unix_ns` - unix timestamp in nanoseconds. Note that VictoriaMetrics rounds the timestamp to milliseconds.
*`rfc3339` - timestamp in [RFC3339](https://tools.ietf.org/html/rfc3339) format, i.e. `2006-01-02T15:04:05Z`.
*`custom:<layout>` - custom layout for the timestamp. The `<layout>` may contain arbitrary time layout according to [time.Parse rules in Go](https://golang.org/pkg/time/#Parse).
Each request to `/api/v1/import/csv` may contain arbitrary number of CSV lines.
Example for importing CSV data via `/api/v1/import/csv`:
Extra labels may be added to all the imported lines by passing `extra_label=name=value` query args.
For example, `/api/v1/import/csv?extra_label=foo=bar` would add `"foo":"bar"` label to all the imported lines.
Note that it could be required to flush response cache after importing historical data. See [these docs](#backfilling) for detail.
### How to import data in Prometheus exposition format
VictoriaMetrics accepts data in [Prometheus exposition format](https://github.com/prometheus/docs/blob/master/content/docs/instrumenting/exposition_formats.md#text-based-format)
and in [OpenMetrics format](https://github.com/OpenObservability/OpenMetrics/blob/master/specification/OpenMetrics.md)
via `/api/v1/import/prometheus` path. For example, the following line imports a single line in Prometheus exposition format into VictoriaMetrics:
Extra labels may be added to all the imported metrics by passing `extra_label=name=value` query args.
For example, `/api/v1/import/prometheus?extra_label=foo=bar` would add `{foo="bar"}` label to all the imported metrics.
If timestamp is missing in `<metric> <value> <timestamp>` Prometheus exposition format line, then the current timestamp is used during data ingestion.
It can be overriden by passing unix timestamp in *milliseconds* via `timestamp` query arg. For example, `/api/v1/import/prometheus?timestamp=1594370496905`.
VictoriaMetrics accepts arbitrary number of lines in a single request to `/api/v1/import/prometheus`, i.e. it supports data streaming.
Note that it could be required to flush response cache after importing historical data. See [these docs](#backfilling) for detail.
VictoriaMetrics also may scrape Prometheus targets - see [these docs](#how-to-scrape-prometheus-exporters-such-as-node-exporter).
## Relabeling
VictoriaMetrics supports Prometheus-compatible relabeling for all the ingested metrics if `-relabelConfig` command-line flag points
to a file containing a list of [relabel_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#relabel_config) entries.
The `-relabelConfig` files can contain special placeholders in the form `%{ENV_VAR}`, which are replaced by the corresponding environment variable values.
VictoriaMetrics provides additional relabeling features such as Graphite-style relabeling. See [these docs](https://docs.victoriametrics.com/vmagent.html#relabeling) for more details.
at `http://<victoriametrics-addr>:8428/federate?match[]=<timeseries_selector_for_federation>`.
Optional `start` and `end` args may be added to the request in order to scrape the last point for each selected time series on the `[start ... end]` interval.
By default, the last point on the interval `[now - max_lookback ... now]` is scraped for each time series. The default value for `max_lookback` is `5m` (5 minutes), but it can be overridden with `max_lookback` query arg.
For instance, `/federate?match[]=up&max_lookback=1h` would return last points on the `[now - 1h ... now]` interval. This may be useful for time series federation
VictoriaMetrics uses lower amounts of CPU, RAM and storage space on production workloads compared to competing solutions (Prometheus, Thanos, Cortex, TimescaleDB, InfluxDB, QuestDB, M3DB) according to [our case studies](https://docs.victoriametrics.com/CaseStudies.html).
VictoriaMetrics capacity scales linearly with the available resources. The needed amounts of CPU and RAM highly depends on the workload - the number of [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series), series [churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate), query types, query qps, etc. It is recommended setting up a test VictoriaMetrics for your production workload and iteratively scaling CPU and RAM resources until it becomes stable according to [troubleshooting docs](#troubleshooting). A single-node VictoriaMetrics works perfectly with the following production workload according to [our case studies](https://docs.victoriametrics.com/CaseStudies.html):
The needed storage space for the given retention (the retention is set via `-retentionPeriod` command-line flag) can be extrapolated from disk space usage in a test run. For example, if `-storageDataPath` directory size becomes 10GB after a day-long test run on a production workload, then it will need at least `10GB*100=1TB` of disk space for `-retentionPeriod=100d` (100-days retention period).
* At least [20% of free storage space](#storage) at the directory pointed by `-storageDataPath` command-line flag. See also `-storage.minFreeDiskSpaceBytes` command-line flag description [here](#list-of-command-line-flags).
See also [resource usage limits docs](#resource-usage-limits).
## Resource usage limits
By default VictoriaMetrics is tuned for an optimal resource usage under typical workloads. Some workloads may need fine-grained resource usage limits. In these cases the following command-line flags may be useful:
-`-memory.allowedPercent` and `-search.allowedBytes` limit the amounts of memory, which may be used for various internal caches at VictoriaMetrics. Note that VictoriaMetrics may use more memory, since these flags don't limit additional memory, which may be needed on a per-query basis.
-`-search.maxUniqueTimeseries` limits the number of unique time series a single query can find and process. VictoriaMetrics keeps in memory some metainformation about the time series located by each query and spends some CPU time for processing the found time series. This means that the maximum memory usage and CPU usage a single query can use is proportional to `-search.maxUniqueTimeseries`.
-`-search.maxQueryDuration` limits the duration of a single query. If the query takes longer than the given duration, then it is canceled. This allows saving CPU and RAM when executing unexpected heavy queries.
-`-search.maxConcurrentRequests` limits the number of concurrent requests VictoriaMetrics can process. Bigger number of concurrent requests usually means bigger memory usage. For example, if a single query needs 100 MiB of additional memory during its execution, then 100 concurrent queries may need `100 * 100 MiB = 10 GiB` of additional memory. So it is better to limit the number of concurrent queries, while suspending additional incoming queries if the concurrency limit is reached. VictoriaMetrics provides `-search.maxQueueDuration` command-line flag for limiting the max wait time for suspended queries.
-`-search.maxSamplesPerSeries` limits the number of raw samples the query can process per each time series. VictoriaMetrics sequentially processes raw samples per each found time series during the query. It unpacks raw samples on the selected time range per each time series into memory and then applies the given [rollup function](https://docs.victoriametrics.com/MetricsQL.html#rollup-functions). The `-search.maxSamplesPerSeries` command-line flag allows limiting memory usage in the case when the query is executed on a time range, which contains hundreds of millions of raw samples per each located time series.
-`-search.maxSamplesPerQuery` limits the number of raw samples a single query can process. This allows limiting CPU usage for heavy queries.
-`-search.maxSeries` limits the number of time series, which may be returned from [/api/v1/series](https://prometheus.io/docs/prometheus/latest/querying/api/#finding-series-by-label-matchers). This endpoint is used mostly by Grafana for auto-completion of metric names, label names and label values. Queries to this endpoint may take big amounts of CPU time and memory when the database contains big number of unique time series because of [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate). In this case it might be useful to set the `-search.maxSeries` to quite low value in order limit CPU and memory usage.
-`-search.maxTagKeys` limits the number of items, which may be returned from [/api/v1/labels](https://prometheus.io/docs/prometheus/latest/querying/api/#getting-label-names). This endpoint is used mostly by Grafana for auto-completion of label names. Queries to this endpoint may take big amounts of CPU time and memory when the database contains big number of unique time series because of [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate). In this case it might be useful to set the `-search.maxTagKeys` to quite low value in order to limit CPU and memory usage.
-`-search.maxTagValues` limits the number of items, which may be returned from [/api/v1/label/.../values](https://prometheus.io/docs/prometheus/latest/querying/api/#querying-label-values). This endpoint is used mostly by Grafana for auto-completion of label values. Queries to this endpoint may take big amounts of CPU time and memory when the database contains big number of unique time series because of [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate). In this case it might be useful to set the `-search.maxTagValues` to quite low value in order to limit CPU and memory usage.
-`-search.maxTagValueSuffixesPerSearch` limits the number of entries, which may be returned from `/metrics/find` endpoint. See [Graphite Metrics API usage docs](#graphite-metrics-api-usage).
VictoriaMetrics leaves a single raw sample with the biggest timestamp per each `-dedup.minScrapeInterval` discrete interval if `-dedup.minScrapeInterval` is set to positive duration. For example, `-dedup.minScrapeInterval=60s` would leave a single raw sample with the biggest timestamp per each discrete 60s interval. If multiple raw samples have the same biggest timestamp on the given `-dedup.minScrapeInterval` discrete interval, then an arbitrary sample out of these samples is left. This aligns with the [staleness rules in Prometheus](https://prometheus.io/docs/prometheus/latest/querying/basics/#staleness).
The `-dedup.minScrapeInterval=D` is equivalent to `-downsampling.period=0s:D` if [downsampling](#downsampling) is enabled. So it is safe to use deduplication and downsampling simultaneously.
The recommended value for `-dedup.minScrapeInterval` must equal to `scrape_interval` config from Prometheus configs. It is recommended to have a single `scrape_interval` across all the scrape targets. See [this article](https://www.robustperception.io/keep-it-simple-scrape_interval-id) for details.
The de-duplication reduces disk space usage if multiple identically configured [vmagent](https://docs.victoriametrics.com/vmagent.html) or Prometheus instances in HA pair
write data to the same VictoriaMetrics instance. These vmagent or Prometheus instances must have identical
`external_labels` section in their configs, so they write data to the same time series. See also [how to set up multiple vmagent instances for scraping the same targets](https://docs.victoriametrics.com/vmagent.html#scraping-big-number-of-targets).
It is recommended passing different `-promscrape.cluster.name` values to HA paris of `vmagent` instances, so the de-duplication consistently leaves samples for one `vmagent` instance and removes duplicate samples from other `vmagent` instances. See [these docs](https://docs.victoriametrics.com/vmagent.html#high-availability) for details.
to a bigger `part` or when the `part` contents goes outside the configured `-retentionPeriod`.
See [this article](https://valyala.medium.com/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282) for more details.
See also [how to work with snapshots](#how-to-work-with-snapshots).
Retention is configured with the `-retentionPeriod` command-line flag, which takes a number followed by a time unit character - `h(ours)`, `d(ays)`, `w(eeks)`, `y(ears)`. If the time unit is not specified, a month is assumed. For instance, `-retentionPeriod=3` means that the data will be stored for 3 months and then deleted. The default retention period is one month.
A single instance of VictoriaMetrics supports only a single retention, which can be configured via `-retentionPeriod` command-line flag. If you need multiple retentions, then you may start multiple VictoriaMetrics instances with distinct values for the following flags:
*`-downsampling.period=30d:5m,180d:1h` instructs VictoriaMetrics to deduplicate samples older than 30 days with 5 minutes interval and to deduplicate samples older than 180 days with 1 hour interval.
Downsampling is applied independently per each time series. It can reduce disk space usage and improve query performance if it is applied to time series with big number of samples per each series. The downsampling doesn't improve query performance if the database contains big number of time series with small number of samples per each series (aka [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate)), since downsampling doesn't reduce the number of time series. So the majority of time is spent on searching for the matching time series. It is possible to use recording rules in [vmalert](https://docs.victoriametrics.com/vmalert.html) in order to reduce the number of time series. See [these docs](https://docs.victoriametrics.com/vmalert.html#downsampling-and-aggregation-via-vmalert).
The downsampling can be evaluated for free by downloading and using enterprise binaries from [the releases page](https://github.com/VictoriaMetrics/VictoriaMetrics/releases).
Single-node VictoriaMetrics doesn't support multi-tenancy. Use the [cluster version](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#multitenancy) instead.
Though single-node VictoriaMetrics cannot scale to multiple nodes, it is optimized for resource usage - storage size / bandwidth / IOPS, RAM, CPU.
This means that a single-node VictoriaMetrics may scale vertically and substitute a moderately sized cluster built with competing solutions
such as Thanos, Uber M3, InfluxDB or TimescaleDB. See [vertical scalability benchmarks](https://medium.com/@valyala/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae).
So try single-node VictoriaMetrics at first and then [switch to the cluster version](https://github.com/VictoriaMetrics/VictoriaMetrics/tree/cluster) if you still need
It is recommended using [vmalert](https://docs.victoriametrics.com/vmalert.html) for alerting.
Additionally, alerting can be set up with the following tools:
* With Prometheus - see [the corresponding docs](https://prometheus.io/docs/alerting/overview/).
* With Promxy - see [the corresponding docs](https://github.com/jacksontj/promxy/blob/master/README.md#how-do-i-use-alertingrecording-rules-in-promxy).
* With Grafana - see [the corresponding docs](https://grafana.com/docs/alerting/rules/).
- All the VictoriaMetrics components must run in protected private networks without direct access from untrusted networks such as Internet. The exception is [vmauth](https://docs.victoriametrics.com/vmauth.html) and [vmgateway](https://docs.victoriametrics.com/vmgateway.html).
- All the requests from untrusted networks to VictoriaMetrics components must go through auth proxy such as vmauth or vmgateway. The proxy must be set up with proper authentication and authorization.
- Prefer using lists of allowed API endpoints, while disallowing access to other endpoints when configuring auth proxy in front of VictoriaMetrics components.
VictoriaMetrics provides the following security-related command-line flags:
For example, substitute `-graphiteListenAddr=:2003` with `-graphiteListenAddr=<internal_iface_ip>:2003`. This protects from unexpected requests from untrusted network interfaces.
The only option is increasing the limit on [the number of open files in the OS](https://medium.com/@muhammadtriwibowo/set-permanently-ulimit-n-open-files-in-ubuntu-4d61064429a).
The recommendation is not specific for VictoriaMetrics only but also for any service which handles many HTTP connections and stores data on disk.
* VictoriaMetrics is a write-heavy application and its performance depends on disk performance. So be careful with other
applications or utilities (like [fstrim](http://manpages.ubuntu.com/manpages/bionic/man8/fstrim.8.html))
which could [exhaust disk resources](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1521).
* The recommended filesystem is `ext4`, the recommended persistent storage is [persistent HDD-based disk on GCP](https://cloud.google.com/compute/docs/disks/#pdspecs),
since it is protected from hardware failures via internal replication and it can be [resized on the fly](https://cloud.google.com/compute/docs/disks/add-persistent-disk#resize_pd).
If you plan to store more than 1TB of data on `ext4` partition or plan extending it to more than 16TB,
then the following options are recommended to pass to `mkfs.ext4`:
VictoriaMetrics returns TSDB stats at `/api/v1/status/tsdb` page in the way similar to Prometheus - see [these Prometheus docs](https://prometheus.io/docs/prometheus/latest/querying/api/#tsdb-stats). VictoriaMetrics accepts the following optional query args at `/api/v1/status/tsdb` page:
*`date=YYYY-MM-DD` where `YYYY-MM-DD` is the date for collecting the stats. By default the stats is collected for the current day. Pass `date=1970-01-01` in order to collect global stats across all the days.
*`focusLabel=LABEL_NAME` returns label values with the highest number of time series for the given `LABEL_NAME` in the `seriesCountByFocusLabelValue` list.
*`match[]=SELECTOR` where `SELECTOR` is an arbitrary [time series selector](https://prometheus.io/docs/prometheus/latest/querying/basics/#time-series-selectors) for series to take into account during stats calculation. By default all the series are taken into account.
*`extra_label=LABEL=VALUE`. See [these docs](#prometheus-querying-api-enhancements) for more details.
*`-storage.maxHourlySeries` - limits the number of time series that can be added during the last hour. Useful for limiting the number of [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series).
*`-storage.maxDailySeries` - limits the number of time series that can be added during the last day. Useful for limiting daily [churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate).
Both limits can be set simultaneously. If any of these limits is reached, then incoming samples for new time series are dropped. A sample of dropped series is put in the log with `WARNING` level.
The exceeded limits can be [monitored](#monitoring) with the following metrics:
*`vm_hourly_series_limit_rows_dropped_total` - the number of metrics dropped due to exceeded hourly limit on the number of unique time series.
*`vm_daily_series_limit_rows_dropped_total` - the number of metrics dropped due to exceeded daily limit on the number of unique time series.
These limits are approximate, so VictoriaMetrics can underflow/overflow the limit by a small percentage (usually less than 1%).
* It is recommended to use default command-line flag values (i.e. don't set them explicitly) until the need
of tweaking these flag values arises.
* It is recommended inspecting logs during troubleshooting, since they may contain useful information.
* It is recommended upgrading to the latest available release from [this page](https://github.com/VictoriaMetrics/VictoriaMetrics/releases),
since the encountered issue could be already fixed there.
* It is recommended to have at least 50% of spare resources for CPU, disk IO and RAM, so VictoriaMetrics could handle short spikes in the workload without performance issues.
* VictoriaMetrics requires free disk space for [merging data files to bigger ones](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282).
It may slow down when there is no enough free space left. So make sure `-storageDataPath` directory
has at least 20% of free space. The remaining amount of free space
can be [monitored](#monitoring) via `vm_free_disk_space_bytes` metric. The total size of data
stored on the disk can be monitored via sum of `vm_data_size_bytes` metrics.
See also `vm_merge_need_free_disk_space` metrics, which are set to values higher than 0
if background merge cannot be initiated due to free disk space shortage. The value shows the number of per-month partitions,
which would start background merge if they had more free disk space.
* VictoriaMetrics buffers incoming data in memory for up to a few seconds before flushing it to persistent storage.
This may lead to the following "issues":
* Data becomes available for querying in a few seconds after inserting. It is possible to flush in-memory buffers to persistent storage
by requesting `/internal/force_flush` http handler. This handler is mostly needed for testing and debugging purposes.
* The last few seconds of inserted data may be lost on unclean shutdown (i.e. OOM, `kill -9` or hardware reset).
See [this article for technical details](https://valyala.medium.com/wal-usage-looks-broken-in-modern-time-series-databases-b62a627ab704).
* If VictoriaMetrics works slowly and eats more than a CPU core per 100K ingested data points per second,
then it is likely you have too many [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series) for the current amount of RAM.
VictoriaMetrics [exposes](#monitoring) `vm_slow_*` metrics such as `vm_slow_row_inserts_total` and `vm_slow_metric_name_loads_total`, which could be used
as an indicator of low amounts of RAM. It is recommended increasing the amount of RAM on the node with VictoriaMetrics in order to improve
ingestion and query performance in this case.
* If the order of labels for the same metrics can change over time (e.g. if `metric{k1="v1",k2="v2"}` may become `metric{k2="v2",k1="v1"}`),
then it is recommended running VictoriaMetrics with `-sortLabels` command-line flag in order to reduce memory usage and CPU usage.
* VictoriaMetrics prioritizes data ingestion over data querying. So if it has no enough resources for data ingestion,
then data querying may slow down significantly.
* If VictoriaMetrics doesn't work because of certain parts are corrupted due to disk errors,
then just remove directories with broken parts. It is safe removing subdirectories under `<-storageDataPath>/data/{big,small}/YYYY_MM` directories
when VictoriaMetrics isn't running. This recovers VictoriaMetrics at the cost of data loss stored in the deleted broken parts.
In the future, `vmrecover` tool will be created for automatic recovering from such errors.
* If you see gaps on the graphs, try resetting the cache by sending request to `/internal/resetRollupResultCache`.
If this removes gaps on the graphs, then it is likely data with timestamps older than `-search.cacheTimestampOffset`
is ingested into VictoriaMetrics. Make sure that data sources have synchronized time with VictoriaMetrics.
If the gaps are related to irregular intervals between samples, then try adjusting `-search.minStalenessInterval` command-line flag
to value close to the maximum interval between samples.
* Metrics and labels leading to [high cardinality](https://docs.victoriametrics.com/FAQ.html#what-is-high-cardinality) or [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate) can be determined via [cardinality explorer](#cardinality-explorer) and via [/api/v1/status/tsdb](#tsdb-stats) endpoint.
* If you store Graphite metrics like `foo.bar.baz` in VictoriaMetrics, then `{__graphite__="foo.*.baz"}` filter can be used for selecting such metrics. See [these docs](#selecting-graphite-metrics) for details.
VictoriaMetrics uses various internal caches. These caches are stored to `<-storageDataPath>/cache` directory during graceful shutdown (e.g. when VictoriaMetrics is stopped by sending `SIGINT` signal). The caches are read on the next VictoriaMetrics startup. Sometimes it is needed to remove such caches on the next startup. This can be performed by placing `reset_cache_on_startup` file inside the `<-storageDataPath>/cache` directory before the restart of VictoriaMetrics. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1447) for details.
should be used only for one-off updates. It shouldn't be used for frequent updates because of non-zero overhead related to data removal.
## Replication
Single-node VictoriaMetrics doesn't support application-level replication. Use cluster version instead.
See [these docs](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#replication-and-data-safety) for details.
Storage-level replication may be offloaded to durable persistent storage such as [Google Cloud disks](https://cloud.google.com/compute/docs/disks#pdspecs).
See also [high availability docs](#high-availability) and [backup docs](#backups).
## Backups
VictoriaMetrics supports backups via [vmbackup](https://docs.victoriametrics.com/vmbackup.html)
and [vmrestore](https://docs.victoriametrics.com/vmrestore.html) tools.
* [Prometheus Oauth proxy](https://gitlab.com/optima_public/prometheus_oauth_proxy) - see [this article](https://medium.com/@richard.holly/powerful-saas-solution-for-detection-metrics-c67b9208d362) for details.
## Contacts
Contact us with any questions regarding VictoriaMetrics at [info@victoriametrics.com](mailto:info@victoriametrics.com).
## Community and contributions
Feel free asking any questions regarding VictoriaMetrics:
[Zip](https://github.com/VictoriaMetrics/VictoriaMetrics/blob/master/VM_logo.zip) contains three folders with different image orientations (main color and inverted version).
Trim timestamps when importing csv data to this duration. Minimum practical duration is 1ms. Higher duration (i.e. 1s) may be used for reducing disk space usage for timestamp data (default 1ms)
Leave only the last sample in every time series per each discrete interval equal to -dedup.minScrapeInterval > 0. See https://docs.victoriametrics.com/#deduplication and https://docs.victoriametrics.com/#downsampling
Whether to deny queries outside of the configured -retentionPeriod. When set, then /api/v1/query_range would return '503 Service Unavailable' error for queries with 'from' value outside -retentionPeriod. This may be useful when multiple data sources with distinct retentions are hidden behind query-tee
Comma-separated downsampling periods in the format 'offset:period'. For example, '30d:10m' instructs to leave a single sample per 10 minutes for samples older than 30 days. See https://docs.victoriametrics.com/#downsampling for details
Supports an array of values separated by comma or specified via multiple flags.
Whether to check only -promscrape.config and then exit. Unknown config entries aren't allowed in -promscrape.config by default. This can be changed with -promscrape.config.strictParse=false command-line flag
Whether to enable reading flags from environment variables additionally to command line. Command line flag values have priority over values from environment vars. Flags are read only from command line if this flag isn't set. See https://docs.victoriametrics.com/#environment-variables for more details
The delay before starting final merge for per-month partition after no new data is ingested into it. Final merge may require additional disk IO and CPU resources. Final merge may increase query speed and reduce disk space usage in some cases. Zero value disables final merge
Whether to use pread() instead of mmap() for reading data files. By default mmap() is used for 64-bit arches and pread() is used for 32-bit arches, since they cannot read data files bigger than 2^32 bytes in memory. mmap() is usually faster for reading small data chunks than pread()
Trim timestamps for Graphite data to this duration. Minimum practical duration is 1s. Higher duration (i.e. 1m) may be used for reducing disk space usage for timestamp data (default 1s)
Incoming http connections are closed after the configured timeout. This may help to spread the incoming load among a cluster of services behind a load balancer. Please note that the real timeout may be bigger by up to 10% as a protection against the thundering herd problem (default 2m0s)
The maximum duration for a graceful shutdown of the HTTP server. A highly loaded server may require increased value for a graceful shutdown (default 7s)
An optional prefix to add to all the paths handled by http server. For example, if '-http.pathPrefix=/foo/bar' is set, then all the http requests will be handled on '/foo/bar/*' paths. This may be useful for proxied requests. See https://www.robustperception.io/using-external-urls-and-proxies-with-prometheus
Optional delay before http server shutdown. During this delay, the server returns non-OK responses from /health page, so load balancers can route new requests to other servers
The maximum length in bytes of a single line accepted by /api/v1/import; the line length can be limited with 'max_rows_per_line' query arg passed to /api/v1/export
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 104857600)
Comma-separated list of database names to return from /query and /influx/query API. This can be needed for accepting data from Telegraf plugins such as https://github.com/fangli/fluent-plugin-influxdb
Supports an array of values separated by comma or specified via multiple flags.
TCP and UDP address to listen for InfluxDB line protocol data. Usually :8089 must be set. Doesn't work if empty. This flag isn't needed when ingesting data over HTTP - just send it to http://<victoriametrics>:8428/write
Trim timestamps for InfluxDB line protocol data to this duration. Minimum practical duration is 1ms. Higher duration (i.e. 1s) may be used for reducing disk space usage for timestamp data (default 1ms)
Whether to log new series. This option is for debug purposes only. It can lead to performance issues when big number of new series are ingested into VictoriaMetrics
Per-second limit on the number of ERROR messages. If more than the given number of errors are emitted per second, the remaining errors are suppressed. Zero values disable the rate limit
Timezone to use for timestamps in logs. Timezone must be a valid IANA Time Zone. For example: America/New_York, Europe/Berlin, Etc/GMT+3 or Local (default "UTC")
Per-second limit on the number of WARN messages. If more than the given number of warns are emitted per second, then the remaining warns are suppressed. Zero values disable the rate limit
The maximum number of concurrent inserts. Default value should work for most cases, since it minimizes the overhead for concurrent inserts. This option is tigthly coupled with -insert.maxQueueDuration (default 16)
The maximum length of label values in the accepted time series. Longer label values are truncated. In this case the vm_too_long_label_values_total metric at /metrics page is incremented (default 16384)
The maximum number of labels accepted per time series. Superfluous labels are dropped. In this case the vm_metrics_with_dropped_labels_total metric at /metrics page is incremented (default 30)
Allowed size of system memory VictoriaMetrics caches may occupy. This option overrides -memory.allowedPercent if set to a non-zero value. Too low a value may increase the cache miss rate usually resulting in higher CPU and disk IO usage. Too high a value may evict too much data from OS page cache resulting in higher disk IO usage
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 0)
Allowed percent of system memory VictoriaMetrics caches may occupy. See also -memory.allowedBytes. Too low a value may increase cache miss rate usually resulting in higher CPU and disk IO usage. Too high a value may evict too much data from OS page cache which will result in higher disk IO usage (default 60)
TCP and UDP address to listen for OpentTSDB metrics. Telnet put messages and HTTP /api/put messages are simultaneously served on TCP port. Usually :4242 must be set. Doesn't work if empty
Trim timestamps for OpenTSDB 'telnet put' data to this duration. Minimum practical duration is 1s. Higher duration (i.e. 1m) may be used for reducing disk space usage for timestamp data (default 1s)
Trim timestamps for OpenTSDB HTTP data to this duration. Minimum practical duration is 1ms. Higher duration (i.e. 1s) may be used for reducing disk space usage for timestamp data (default 1ms)
The number of number in the cluster of scrapers. It must be an unique value in the range 0 ... promscrape.cluster.membersCount-1 across scrapers in the cluster. Can be specified as pod name of Kubernetes StatefulSet - pod-name-Num, where Num is a numeric part of pod name (default "0")
The number of members in a cluster of scrapers. Each member must have an unique -promscrape.cluster.memberNum in the range 0 ... promscrape.cluster.membersCount-1 . Each member then scrapes roughly 1/N of all the targets. By default cluster scraping is disabled, i.e. a single scraper scrapes all the targets
Optional name of the cluster. If multiple vmagent clusters scrape the same targets, then each cluster must have unique name in order to properly de-duplicate samples received from these clusters. See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2679
The number of members in the cluster, which scrape the same targets. If the replication factor is greater than 1, then the deduplication must be enabled at remote storage side. See https://docs.victoriametrics.com/#deduplication (default 1)
Optional path to Prometheus config file with 'scrape_configs' section containing targets to scrape. The path can point to local file and to http url. See https://docs.victoriametrics.com/#how-to-scrape-prometheus-exporters-such-as-node-exporter for details
Checks -promscrape.config file for errors and unsupported fields and then exits. Returns non-zero exit code on parsing errors and emits these errors to stderr. See also -promscrape.config.strictParse command-line flag. Pass -loggerLevel=ERROR if you don't need to see info messages in the output.
Interval for checking for changes in '-promscrape.config' file. By default the checking is disabled. Send SIGHUP signal in order to force config check for changes
Interval for checking for changes in Consul. This works only if consul_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#consul_sd_config for details (default 30s)
Interval for checking for changes in digital ocean. This works only if digitalocean_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#digitalocean_sd_config for details (default 1m0s)
Whether to disable sending 'Accept-Encoding: gzip' request headers to all the scrape targets. This may reduce CPU usage on scrape targets at the cost of higher network bandwidth utilization. It is possible to set 'disable_compression: true' individually per each 'scrape_config' section in '-promscrape.config' for fine grained control
Whether to disable HTTP keep-alive connections when scraping all the targets. This may be useful when targets has no support for HTTP keep-alive connection. It is possible to set 'disable_keepalive: true' individually per each 'scrape_config' section in '-promscrape.config' for fine grained control. Note that disabling HTTP keep-alive may increase load on both vmagent and scrape targets
The maximum duration for waiting to perform API requests if more than -promscrape.discovery.concurrency requests are simultaneously performed (default 1m0s)
Interval for checking for changes in dns. This works only if dns_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#dns_sd_config for details (default 30s)
Interval for checking for changes in docker. This works only if docker_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#docker_sd_config for details (default 30s)
Interval for checking for changes in dockerswarm. This works only if dockerswarm_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#dockerswarm_sd_config for details (default 30s)
Whether to drop original labels for scrape targets at /targets and /api/v1/targets pages. This may be needed for reducing memory usage when original labels for big number of scrape targets occupy big amounts of memory. Note that this reduces debuggability for improper per-target relabeling configs
Interval for checking for changes in ec2. This works only if ec2_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#ec2_sd_config for details (default 1m0s)
Interval for checking for changes in eureka. This works only if eureka_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#eureka_sd_config for details (default 30s)
Interval for checking for changes in 'file_sd_config'. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#file_sd_config for details (default 5m0s)
Interval for checking for changes in gce. This works only if gce_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#gce_sd_config for details (default 1m0s)
Interval for checking for changes in http endpoint service discovery. This works only if http_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#http_sd_config for details (default 1m0s)
Interval for checking for changes in Kubernetes API server. This works only if kubernetes_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#kubernetes_sd_config for details (default 30s)
The maximum number of droppedTargets to show at /api/v1/targets page. Increase this value if your setup drops more scrape targets during relabeling and you need investigating labels for all the dropped targets. Note that the increased number of tracked dropped targets may result in increased memory usage (default 1000)
The minimum target response size for automatic switching to stream parsing mode, which can reduce memory usage. See https://docs.victoriametrics.com/vmagent.html#stream-parsing-mode
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 1000000)
Whether to disable sending Prometheus stale markers for metrics when scrape target disappears. This option may reduce memory usage if stale markers aren't needed for your setup. This option also disables populating the scrape_series_added metric. See https://prometheus.io/docs/concepts/jobs_instances/#automatically-generated-labels-and-time-series
Interval for checking for changes in openstack API server. This works only if openstack_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#openstack_sd_config for details (default 30s)
Optional limit on the number of unique time series a single scrape target can expose. See https://docs.victoriametrics.com/vmagent.html#cardinality-limiter for more info
Whether to enable stream parsing for metrics obtained from scrape targets. This may be useful for reducing memory usage when millions of metrics are exposed per each scrape target. It is posible to set 'stream_parse: true' individually per each 'scrape_config' section in '-promscrape.config' for fine grained control
Whether to suppress scrape errors logging. The last error for each target is always available at '/targets' page even if scrape errors logging is suppressed. See also -promscrape.suppressScrapeErrorsDelay
-promscrape.suppressScrapeErrorsDelay duration
The delay for suppressing repeated scrape errors logging per each scrape targets. This may be used for reducing the number of log lines related to scrape errors. See also -promscrape.suppressScrapeErrors
Optional path to a file with relabeling rules, which are applied to all the ingested metrics. The path can point either to local file or to http url. See https://docs.victoriametrics.com/#relabeling for details. The config is reloaded on SIGHUP signal
Whether to log metrics before and after relabeling with -relabelConfig. If the -relabelDebug is enabled, then the metrics aren't sent to storage. This is useful for debugging the relabeling configs
Data with timestamps outside the retentionPeriod is automatically deleted
The following optional suffixes are supported: h (hour), d (day), w (week), y (year). If suffix isn't set, then the duration is counted in months (default 1)
The offset for performing indexdb rotation. If set to 0, then the indexdb rotation is performed at 4am UTC time per each -retentionPeriod. If set to 2h, then the indexdb rotation is performed at 4am EET time (the timezone with +2h offset)
The maximum duration since the current time for response data, which is always queried from the original raw data, without using the response cache. Increase this value if you see gaps in responses due to time synchronization issues between VictoriaMetrics and data sources. See also -search.disableAutoCacheReset (default 5m0s)
The interval between datapoints stored in the database. It is used at Graphite Render API handler for normalizing the interval between datapoints in case it isn't normalized. It can be overriden by sending 'storage_step' query arg to /render API or by sending the desired interval via 'Storage-Step' http header during querying /render API (default 10s)
The time when data points become visible in query results after the collection. Too small value can result in incomplete last points for query results (default 30s)
The maximum number of concurrent search requests. It shouldn't be high, since a single request can saturate all the CPU cores. See also -search.maxQueueDuration (default 8)
The maximum number of time series, which can be scanned during queries to Graphite Render API. See https://docs.victoriametrics.com/#graphite-render-api-usage (default 300000)
Synonym to -search.lookback-delta from Prometheus. The value is dynamically detected from interval between time series datapoints if not set. It can be overridden on per-query basis via max_lookback arg. See also '-search.maxStalenessInterval' flag, which has the same meaining due to historical reasons
The maximum points per a single timeseries returned from /api/v1/query_range. This option doesn't limit the number of scanned raw samples in the database. The main purpose of this option is to limit the number of per-series points returned to graphing UI such as Grafana. There is no sense in setting this limit to values bigger than the horizontal resolution of the graph (default 30000)
The maximum number of raw samples a single query can process across all time series. This protects from heavy queries, which select unexpectedly high number of raw samples. See also -search.maxSamplesPerSeries (default 1000000000)
The maximum interval for staleness calculations. By default it is automatically calculated from the median interval between samples. This flag could be useful for tuning Prometheus data model closer to Influx-style data model. See https://prometheus.io/docs/prometheus/latest/querying/basics/#staleness for details. See also '-search.setLookbackToStep' flag
The maximum step when /api/v1/query_range handler adjusts points with timestamps closer than -search.latencyOffset to the current time. The adjustment is needed because such points may contain incomplete data (default 1m0s)
The maximum number of time series, which can be processed during the call to /api/v1/status/tsdb. This option allows limiting memory usage (default 10000000)
The maximum number of unique time series, which can be selected during /api/v1/query and /api/v1/query_range queries. This option allows limiting memory usage (default 300000)
The minimum interval for staleness calculations. This flag could be useful for removing gaps on graphs generated from time series with irregular intervals between samples. See also '-search.maxStalenessInterval'
Set this flag to true if the database doesn't contain Prometheus stale markers, so there is no need in spending additional CPU time on its handling. Staleness markers may exist only in data obtained from Prometheus scrape targets
The minimum duration for queries to track in query stats at /api/v1/status/top_queries. Queries with lower duration are ignored in query stats (default 1ms)
Whether to fix lookback interval to 'step' query arg value. If set to true, the query model becomes closer to InfluxDB data model. If set to true, then -search.maxLookback and -search.maxStalenessInterval are ignored
Whether to treat dots as is in regexp label filters used in queries. For example, foo{bar=~"a.b.c"} will be automatically converted to foo{bar=~"a\\.b\\.c"}, i.e. all the dots in regexp filters will be automatically escaped in order to match only dot char instead of matching any char. Dots in ".+", ".*" and ".{n}" regexps aren't escaped. This option is DEPRECATED in favor of {__graphite__="a.*.c"} syntax for selecting metrics matching the given Graphite metrics filter
Automatically delete snapshots older than -snapshotsMaxAge if it is set to non-zero duration. Make sure that backup process has enough time to finish the backup before the corresponding snapshot is automatically deleted
The following optional suffixes are supported: h (hour), d (day), w (week), y (year). If suffix isn't set, then the duration is counted in months (default 0)
Whether to sort labels for incoming samples before writing them to storage. This may be needed for reducing memory usage at storage when the order of labels in incoming samples is random. For example, if m{k1="v1",k2="v2"} may be sent as m{k2="v2",k1="v1"}. Enabled sorting for labels can slow down ingestion performance a bit
The maximum number of unique series can be added to the storage during the last 24 hours. Excess series are logged and dropped. This can be useful for limiting series churn rate. See also -storage.maxHourlySeries
The maximum number of unique series can be added to the storage during the last hour. Excess series are logged and dropped. This can be useful for limiting series cardinality. See also -storage.maxDailySeries
Path to file with TLS certificate if -tls is set. Prefer ECDSA certs instead of RSA certs as RSA certs are slower. The provided certificate file is automatically re-read every second, so it can be dynamically updated
Optional list of TLS cipher suites for incoming requests over HTTPS if -tls is set. See the list of supported cipher suites at https://pkg.go.dev/crypto/tls#pkg-constants
Optional URL for proxying requests to vmalert. For example, if -vmalert.proxyURL=http://vmalert:8880 , then alerting API requests such as /api/v1/rules from Grafana will be proxied to http://vmalert:8880/api/v1/rules
