VictoriaMetrics/docs/VictoriaLogs/data-ingestion/README.md

267 lines
16 KiB
Markdown
Raw Normal View History

---
sort: 3
title: Data ingestion
weight: 3
menu:
docs:
identifier: victorialogs-data-ingestion
parent: "victorialogs"
weight: 3
aliases:
- /VictoriaLogs/data-ingestion/
- /VictoriaLogs/data-ingestion/index.html
---
# Data ingestion
2024-05-24 22:30:58 +00:00
[VictoriaLogs](https://docs.victoriametrics.com/victorialogs/) can accept logs from the following log collectors:
2024-06-17 10:13:18 +00:00
- Syslog - see [these docs](https://docs.victoriametrics.com/victorialogs/data-ingestion/syslog/).
- Filebeat - see [these docs](https://docs.victoriametrics.com/victorialogs/data-ingestion/filebeat/).
- Fluentbit - see [these docs](https://docs.victoriametrics.com/victorialogs/data-ingestion/fluentbit/).
- Logstash - see [these docs](https://docs.victoriametrics.com/victorialogs/data-ingestion/logstash/).
- Vector - see [these docs](https://docs.victoriametrics.com/victorialogs/data-ingestion/vector/).
- Promtail (aka Grafana Loki) - see [these docs](https://docs.victoriametrics.com/victorialogs/data-ingestion/promtail/).
2024-05-24 22:30:58 +00:00
The ingested logs can be queried according to [these docs](https://docs.victoriametrics.com/victorialogs/querying/).
See also:
- [Log collectors and data ingestion formats](#log-collectors-and-data-ingestion-formats).
- [Data ingestion troubleshooting](#troubleshooting).
## HTTP APIs
VictoriaLogs supports the following data ingestion HTTP APIs:
- Elasticsearch bulk API. See [these docs](#elasticsearch-bulk-api).
- JSON stream API aka [ndjson](https://jsonlines.org/). See [these docs](#json-stream-api).
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
- Loki JSON API. See [these docs](#loki-json-api).
VictoriaLogs accepts optional [HTTP parameters](#http-parameters) at data ingestion HTTP APIs.
### Elasticsearch bulk API
VictoriaLogs accepts logs in [Elasticsearch bulk API](https://www.elastic.co/guide/en/elasticsearch/reference/current/docs-bulk.html)
/ [OpenSearch Bulk API](http://opensearch.org/docs/1.2/opensearch/rest-api/document-apis/bulk/) format
at `http://localhost:9428/insert/elasticsearch/_bulk` endpoint.
The following command pushes a single log line to VictoriaLogs:
```sh
echo '{"create":{}}
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
{"_msg":"cannot open file","_time":"0","host.name":"host123"}
' | curl -X POST -H 'Content-Type: application/json' --data-binary @- http://localhost:9428/insert/elasticsearch/_bulk
```
It is possible to push thousands of log lines in a single request to this API.
2024-05-24 22:30:58 +00:00
If the [timestamp field](https://docs.victoriametrics.com/victorialogs/keyconcepts/#time-field) is set to `"0"`,
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
then the current timestamp at VictoriaLogs side is used per each ingested log line.
Otherwise the timestamp field must be in the [ISO8601](https://en.wikipedia.org/wiki/ISO_8601) format. For example, `2023-06-20T15:32:10Z`.
Optional fractional part of seconds can be specified after the dot - `2023-06-20T15:32:10.123Z`.
Timezone can be specified instead of `Z` suffix - `2023-06-20T15:32:10+02:00`.
2024-05-24 22:30:58 +00:00
See [these docs](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model) for details on fields,
which must be present in the ingested log messages.
The API accepts various http parameters, which can change the data ingestion behavior - [these docs](#http-parameters) for details.
2024-05-24 22:30:58 +00:00
The following command verifies that the data has been successfully ingested to VictoriaLogs by [querying](https://docs.victoriametrics.com/victorialogs/querying/) it:
```sh
curl http://localhost:9428/select/logsql/query -d 'query=host.name:host123'
```
The command should return the following response:
```sh
{"_msg":"cannot open file","_stream":"{}","_time":"2023-06-21T04:24:24Z","host.name":"host123"}
```
The response by default contains all the [log fields](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model).
2024-05-24 22:30:58 +00:00
See [how to query specific fields](https://docs.victoriametrics.com/victorialogs/logsql/#querying-specific-fields).
The duration of requests to `/insert/elasticsearch/_bulk` can be monitored with `vl_http_request_duration_seconds{path="/insert/elasticsearch/_bulk"}` metric.
See also:
- [How to debug data ingestion](#troubleshooting).
- [HTTP parameters, which can be passed to the API](#http-parameters).
2024-05-24 22:30:58 +00:00
- [How to query VictoriaLogs](https://docs.victoriametrics.com/victorialogs/querying/).
### JSON stream API
VictoriaLogs accepts JSON line stream aka [ndjson](https://jsonlines.org/) at `http://localhost:9428/insert/jsonline` endpoint.
The following command pushes multiple log lines to VictoriaLogs:
```sh
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
echo '{ "log": { "level": "info", "message": "hello world" }, "date": "0", "stream": "stream1" }
{ "log": { "level": "error", "message": "oh no!" }, "date": "0", "stream": "stream1" }
{ "log": { "level": "info", "message": "hello world" }, "date": "0", "stream": "stream2" }
' | curl -X POST -H 'Content-Type: application/stream+json' --data-binary @- \
'http://localhost:9428/insert/jsonline?_stream_fields=stream&_time_field=date&_msg_field=log.message'
```
It is possible to push unlimited number of log lines in a single request to this API.
2024-05-24 22:30:58 +00:00
If the [timestamp field](https://docs.victoriametrics.com/victorialogs/keyconcepts/#time-field) is set to `"0"`,
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
then the current timestamp at VictoriaLogs side is used per each ingested log line.
Otherwise the timestamp field must be in the [ISO8601](https://en.wikipedia.org/wiki/ISO_8601) format. For example, `2023-06-20T15:32:10Z`.
Optional fractional part of seconds can be specified after the dot - `2023-06-20T15:32:10.123Z`.
Timezone can be specified instead of `Z` suffix - `2023-06-20T15:32:10+02:00`.
2024-05-24 22:30:58 +00:00
See [these docs](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model) for details on fields,
which must be present in the ingested log messages.
The API accepts various http parameters, which can change the data ingestion behavior - [these docs](#http-parameters) for details.
2024-05-24 22:30:58 +00:00
The following command verifies that the data has been successfully ingested into VictoriaLogs by [querying](https://docs.victoriametrics.com/victorialogs/querying/) it:
```sh
curl http://localhost:9428/select/logsql/query -d 'query=log.level:*'
```
The command should return the following response:
```sh
{"_msg":"hello world","_stream":"{stream=\"stream2\"}","_time":"2023-06-20T13:35:11.56789Z","log.level":"info"}
{"_msg":"hello world","_stream":"{stream=\"stream1\"}","_time":"2023-06-20T15:31:23Z","log.level":"info"}
{"_msg":"oh no!","_stream":"{stream=\"stream1\"}","_time":"2023-06-20T15:32:10.567Z","log.level":"error"}
```
The response by default contains all the [log fields](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model).
2024-05-24 22:30:58 +00:00
See [how to query specific fields](https://docs.victoriametrics.com/victorialogs/logsql/#querying-specific-fields).
The duration of requests to `/insert/jsonline` can be monitored with `vl_http_request_duration_seconds{path="/insert/jsonline"}` metric.
See also:
- [How to debug data ingestion](#troubleshooting).
- [HTTP parameters, which can be passed to the API](#http-parameters).
2024-05-24 22:30:58 +00:00
- [How to query VictoriaLogs](https://docs.victoriametrics.com/victorialogs/querying/).
app/vlinsert: add support of loki push protocol (#4482) * app/vlinsert: add support of loki push protocol - implemented loki push protocol for both Protobuf and JSON formats - added examples in documentation - added example docker-compose Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert: move protobuf metric into its own file Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: update reference to docker image Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: make volume name unique Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: add license reference Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: fix volume name Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * docs/VictoriaLogs/data-ingestion: add stream fields for loki JSON ingestion example Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: move entities to places where those are used Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: refactor to use common components - use CommonParameters from insertutils - stop ingestion after first error similar to elasticsearch and jsonline Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: address review feedback - add missing logstorage.PutLogRows calls - refactor tenant ID parsing to use common function - reduce number of allocations for parsing by reusing logfields slices - add tests and benchmarks for requests processing funcs Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> --------- Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
2023-07-20 08:10:55 +00:00
### Loki JSON API
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
VictoriaLogs accepts logs in [Loki JSON API](https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki) format at `http://localhost:9428/insert/loki/api/v1/push` endpoint.
app/vlinsert: add support of loki push protocol (#4482) * app/vlinsert: add support of loki push protocol - implemented loki push protocol for both Protobuf and JSON formats - added examples in documentation - added example docker-compose Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert: move protobuf metric into its own file Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: update reference to docker image Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: make volume name unique Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: add license reference Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: fix volume name Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * docs/VictoriaLogs/data-ingestion: add stream fields for loki JSON ingestion example Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: move entities to places where those are used Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: refactor to use common components - use CommonParameters from insertutils - stop ingestion after first error similar to elasticsearch and jsonline Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: address review feedback - add missing logstorage.PutLogRows calls - refactor tenant ID parsing to use common function - reduce number of allocations for parsing by reusing logfields slices - add tests and benchmarks for requests processing funcs Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> --------- Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
2023-07-20 08:10:55 +00:00
The following command pushes a single log line to Loki JSON API at VictoriaLogs:
```sh
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
curl -H "Content-Type: application/json" -XPOST "http://localhost:9428/insert/loki/api/v1/push?_stream_fields=instance,job" --data-raw \
'{"streams": [{ "stream": { "instance": "host123", "job": "app42" }, "values": [ [ "0", "foo fizzbuzz bar" ] ] }]}'
```
It is possible to push thousands of log streams and log lines in a single request to this API.
The API accepts various http parameters, which can change the data ingestion behavior - [these docs](#http-parameters) for details.
There is no need in specifying `_msg_field` and `_time_field` query args, since VictoriaLogs automatically extracts log message and timestamp from the ingested Loki data.
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
2024-05-24 22:30:58 +00:00
The following command verifies that the data has been successfully ingested into VictoriaLogs by [querying](https://docs.victoriametrics.com/victorialogs/querying/) it:
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
```sh
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
curl http://localhost:9428/select/logsql/query -d 'query=fizzbuzz'
app/vlinsert: add support of loki push protocol (#4482) * app/vlinsert: add support of loki push protocol - implemented loki push protocol for both Protobuf and JSON formats - added examples in documentation - added example docker-compose Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert: move protobuf metric into its own file Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: update reference to docker image Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: make volume name unique Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: add license reference Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * deployment/docker/victorialogs/promtail: fix volume name Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * docs/VictoriaLogs/data-ingestion: add stream fields for loki JSON ingestion example Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: move entities to places where those are used Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: refactor to use common components - use CommonParameters from insertutils - stop ingestion after first error similar to elasticsearch and jsonline Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> * app/vlinsert/loki: address review feedback - add missing logstorage.PutLogRows calls - refactor tenant ID parsing to use common function - reduce number of allocations for parsing by reusing logfields slices - add tests and benchmarks for requests processing funcs Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com> --------- Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
2023-07-20 08:10:55 +00:00
```
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
The command should return the following response:
```sh
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
{"_msg":"foo fizzbuzz bar","_stream":"{instance=\"host123\",job=\"app42\"}","_time":"2023-07-20T23:01:19.288676497Z"}
```
The response by default contains all the [log fields](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model).
2024-05-24 22:30:58 +00:00
See [how to query specific fields](https://docs.victoriametrics.com/victorialogs/logsql/#querying-specific-fields).
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
The duration of requests to `/insert/loki/api/v1/push` can be monitored with `vl_http_request_duration_seconds{path="/insert/loki/api/v1/push"}` metric.
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
See also:
- [How to debug data ingestion](#troubleshooting).
- [HTTP parameters, which can be passed to the API](#http-parameters).
2024-05-24 22:30:58 +00:00
- [How to query VictoriaLogs](https://docs.victoriametrics.com/victorialogs/querying/).
app/vlinsert/loki: follow-up after 09df5b66fd7bf33f171d6179748e6f32394ef56e - Parse protobuf if Content-Type isn't set to `application/json` - this behavior is documented at https://grafana.com/docs/loki/latest/api/#push-log-entries-to-loki - Properly handle gzip'ped JSON requests. The `gzip` header must be read from `Content-Encoding` instead of `Content-Type` header - Properly flush all the parsed logs with the explicit call to vlstorage.MustAddRows() at the end of query handler - Check JSON field types more strictly. - Allow parsing Loki timestamp as floating-point number. Such a timestamp can be generated by some clients, which store timestamps in float64 instead of int64. - Optimize parsing of Loki labels in Prometheus text exposition format. - Simplify tests. - Remove lib/slicesutil, since there are no more users for it. - Update docs with missing info and fix various typos. For example, it should be enough to have `instance` and `job` labels as stream fields in most Loki setups. - Allow empty of missing timestamps in the ingested logs. The current timestamp at VictoriaLogs side is then used for the ingested logs. This simplifies debugging and testing of the provided HTTP-based data ingestion APIs. The remaining MAJOR issue, which needs to be addressed: victoria-logs binary size increased from 13MB to 22MB after adding support for Loki data ingestion protocol at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4482 . This is because of shitty protobuf dependencies. They must be replaced with another protobuf implementation similar to the one used at lib/prompb or lib/prompbmarshal .
2023-07-20 23:21:47 +00:00
### HTTP parameters
VictoriaLogs accepts the following parameters at [data ingestion HTTP APIs](#http-apis):
2024-05-24 22:30:58 +00:00
- `_msg_field` - it must contain the name of the [log field](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model)
with the [log message](https://docs.victoriametrics.com/victorialogs/keyconcepts/#message-field) generated by the log shipper.
This is usually the `message` field for Filebeat and Logstash.
If the `_msg_field` parameter isn't set, then VictoriaLogs reads the log message from the `_msg` field.
2024-05-24 22:30:58 +00:00
- `_time_field` - it must contain the name of the [log field](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model)
with the [log timestamp](https://docs.victoriametrics.com/victorialogs/keyconcepts/#time-field) generated by the log shipper.
This is usually the `@timestamp` field for Filebeat and Logstash.
If the `_time_field` parameter isn't set, then VictoriaLogs reads the timestamp from the `_time` field.
If this field doesn't exist, then the current timestamp is used.
2024-05-24 22:30:58 +00:00
- `_stream_fields` - it should contain comma-separated list of [log field](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model) names,
which uniquely identify every [log stream](https://docs.victoriametrics.com/victorialogs/keyconcepts/#stream-fields) collected the log shipper.
If the `_stream_fields` parameter isn't set, then all the ingested logs are written to default log stream - `{}`.
2024-05-24 22:30:58 +00:00
- `ignore_fields` - this parameter may contain the list of [log field](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model) names,
which must be ignored during data ingestion.
- `debug` - if this parameter is set to `1`, then the ingested logs aren't stored in VictoriaLogs. Instead,
the ingested data is logged by VictoriaLogs, so it can be investigated later.
See also [HTTP headers](#http-headers).
### HTTP headers
VictoriaLogs accepts optional `AccountID` and `ProjectID` headers at [data ingestion HTTP APIs](#http-apis).
2024-05-24 22:30:58 +00:00
These headers may contain the needed tenant to ingest data to. See [multitenancy docs](https://docs.victoriametrics.com/victorialogs/#multitenancy) for details.
## Troubleshooting
The following command can be used for verifying whether the data is successfully ingested into VictoriaLogs:
```sh
curl http://localhost:9428/select/logsql/query -d 'query=*' | head
```
2024-05-24 22:30:58 +00:00
This command selects all the data ingested into VictoriaLogs via [HTTP query API](https://docs.victoriametrics.com/victorialogs/querying/#http-api)
using [any value filter](https://docs.victoriametrics.com/victorialogs/logsql/#any-value-filter),
while `head` cancels query execution after reading the first 10 log lines. See [these docs](https://docs.victoriametrics.com/victorialogs/querying/#command-line)
for more details on how `head` integrates with VictoriaLogs.
The response by default contains all the [log fields](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model).
2024-05-24 22:30:58 +00:00
See [how to query specific fields](https://docs.victoriametrics.com/victorialogs/logsql/#querying-specific-fields).
VictoriaLogs provides the following command-line flags, which can help debugging data ingestion issues:
- `-logNewStreams` - if this flag is passed to VictoriaLogs, then it logs all the newly
2024-05-24 22:30:58 +00:00
registered [log streams](https://docs.victoriametrics.com/victorialogs/keyconcepts/#stream-fields).
This may help debugging [high cardinality issues](https://docs.victoriametrics.com/victorialogs/keyconcepts/#high-cardinality).
- `-logIngestedRows` - if this flag is passed to VictoriaLogs, then it logs all the ingested
2024-05-24 22:30:58 +00:00
[log entries](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model).
See also `debug` [parameter](#http-parameters).
2024-05-24 22:30:58 +00:00
VictoriaLogs exposes various [metrics](https://docs.victoriametrics.com/victorialogs/#monitoring), which may help debugging data ingestion issues:
2024-05-24 22:30:58 +00:00
- `vl_rows_ingested_total` - the number of ingested [log entries](https://docs.victoriametrics.com/victorialogs/keyconcepts/#data-model)
since the last VictoriaLogs restart. If this number increases over time, then logs are successfully ingested into VictoriaLogs.
The ingested logs can be inspected in the following ways:
- By passing `debug=1` parameter to every request to [data ingestion APIs](#http-apis). The ingested rows aren't stored in VictoriaLogs
in this case. Instead, they are logged, so they can be investigated later.
2024-05-24 22:30:58 +00:00
The `vl_rows_dropped_total` [metric](https://docs.victoriametrics.com/victorialogs/#monitoring) is incremented for each logged row.
- By passing `-logIngestedRows` command-line flag to VictoriaLogs. In this case it logs all the ingested data, so it can be investigated later.
2024-05-24 22:30:58 +00:00
- `vl_streams_created_total` - the number of created [log streams](https://docs.victoriametrics.com/victorialogs/keyconcepts/#stream-fields)
since the last VictoriaLogs restart. If this metric grows rapidly during extended periods of time, then this may lead
2024-05-24 22:30:58 +00:00
to [high cardinality issues](https://docs.victoriametrics.com/victorialogs/keyconcepts/#high-cardinality).
The newly created log streams can be inspected in logs by passing `-logNewStreams` command-line flag to VictoriaLogs.
## Log collectors and data ingestion formats
Here is the list of log collectors and their ingestion formats supported by VictoriaLogs:
| How to setup the collector | Format: Elasticsearch | Format: JSON Stream | Format: Loki |
|------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------|---------------------------------------------------------------|-------------------------------------------------------------------------------------|
2024-06-17 10:13:18 +00:00
| [Filebeat](https://docs.victoriametrics.com/victorialogs/data-ingestion/filebeat/) | [Yes](https://www.elastic.co/guide/en/beats/filebeat/current/elasticsearch-output.html) | No | No |
| [Fluentbit](https://docs.victoriametrics.com/victorialogs/data-ingestion/fluentbit/) | No | [Yes](https://docs.fluentbit.io/manual/pipeline/outputs/http) | [Yes](https://docs.fluentbit.io/manual/pipeline/outputs/loki) |
| [Logstash](https://docs.victoriametrics.com/victorialogs/data-ingestion/logstash/) | [Yes](https://www.elastic.co/guide/en/logstash/current/plugins-outputs-elasticsearch.html) | No | No |
| [Vector](https://docs.victoriametrics.com/victorialogs/data-ingestion/vector/) | [Yes](https://vector.dev/docs/reference/configuration/sinks/elasticsearch/) | [Yes](https://vector.dev/docs/reference/configuration/sinks/http/) | [Yes](https://vector.dev/docs/reference/configuration/sinks/loki/) |
| [Promtail](https://docs.victoriametrics.com/victorialogs/data-ingestion/promtail/) | No | No | [Yes](https://grafana.com/docs/loki/latest/clients/promtail/configuration/#clients) |