This reduces the number of memory allocations at the cost of possible memory usage increase,
since now different metric name strings may hold references to the previous byte slice.
This is good tradeoff, since ProcessSearchQuery is called in vmselect, and vmselect isn't usually limited by memory.
This change has been extracted from https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5527
* app/vmselect: limit the number of parallel workers by 32
The change should improve performance and memory usage during query processing
on machines with big number of CPU cores. The number of parallel workers for
query processing is controlled via `-search.maxWorkersPerQuery` command-line flag.
By default, the number of workers is limited by the number of available CPU cores,
but not more than 32. The limit can be increased via `-search.maxWorkersPerQuery`.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* wip
- The `-search.maxWorkersPerQuery` command-line flag doesn't limit resource usage,
so move it from the `resource usage limits` to `troubleshooting` chapter at docs/Single-server-VictoriaMetrics.md
- Make more clear the description for the `-search.maxWorkersPerQuery` command-line flag
- Add the description of `-search.maxWorkersPerQuery` to docs/Cluster-VictoriaMetrics.md
- Limit the maximum value, which can be passed to `-search.maxWorkersPerQuery`, to GOMAXPROCS,
because bigger values may worsen query performance and increase CPU usage
- Improve the the description of the change at docs/CHANGELOG.md. Mark it as FEATURE instead of BUGFIX,
since it is closer to a feature than to a bugfix.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5087
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
- Document the change at docs/CHANGELOG.md
- Set the default value for -vmstorageUserTimeout to 3 seconds. This is much better
than the 0 value, which means that TCP connection to unreachable vmstorage could block
for up to 16 minutes.
- Document -vmstorageUserTimeout at docs/Cluster-VictoriaMetrics.md
`TCP_USER_TIMEOUT` (since Linux 2.6.37) specifies the maximum amount of
time that transmitted data may remain unacknowledged before TCP will
forcibly close the connection and return `ETIMEDOUT` to the application.
Setting a low TCP user timeout allows RPC connections quickly reroute
around unavailable storage nodes during network interruptions.
This eliminates the need in .(*T) casting for results obtained from Load()
Leave atomic.Value for map, since atomic.Pointer[map[...]...] makes double pointer to map,
because map is already a pointer type.
- Clarify docs about -replicationFactor command-line flag at vmselect
- Clarify description for -replicationFactor and -search.skipSlowReplicas command-line flags
- Fix the logic for returning responses if -search.skipSlowReplicas command-line flag
is enabled. The logic was broken in the 173ccf4333,
so it could return responses only if some of vmstorage nodes return error,
while it should return when query results are successfully collected from more than
(len(storageNodes) - replicationFactor) vmstorage nodes.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1207
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711
* vmselect: introduce `search.skipSlowReplicas` cmd-line flag
vmselect has two logical conditions during request processing when
`-replicationFactor` cmd-line flag is set:
1. If at least `len(storageNodes) - replicationFactor` responded, it could skip
waiting for the rest of nodes to respond. This could lead to problems described
here https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1207.
2. Mark response as partial if less than `len(storageNodes) - replicationFactor` responded
without an error.
The P1 showed itself error-prone and became the main reason why
`-replicationFactor` wasn't recommended to use at vmselect level.
However, this optimization could be still very useful in situations
when there are slow and fast replicas in cluster.
But P2 remains viable and important conditionless.
Hiding P1 behind the feature-flag `search.skipSlowReplicas`
should make `-replicationFactor` flag usable again. And let users
choose whether they want P1 to be respected.
Related issues
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1207https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* docs: update changelog
Signed-off-by: hagen1778 <roman@victoriametrics.com>
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
- Clarify the scope of the fix at docs/CHANGELOG.md
- Handle the case when -search.maxSamplesPerSeries limit is exceeded
in the same way as the -search.maxSamplesPerQuery limit.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/4472
Properly return the error to user when `-search.maxSamplesPerQuery` limit is exceeded.
Before, user could have received a partial response instead.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Callers of these functions log the returned error and then exit. The returned error already contains the path
to directory, which was failed to be created. So let's just log the error together with the call stack
inside these functions. This leaves the debuggability of the returned error at the same level
while allows simplifying the code at callers' side.
While at it, properly use MustMkdirFailIfExist instead of MustMkdirIfNotExist inside inmemoryPart.MustStoreToDisk().
It is expected that the inmemoryPart.MustStoreToDick() must fail if there is already a directory under the given path.
using `runtime.Gosched` requires acquiring global lock to check if there are any other goroutines to perform tasks. with the latest versions of runtime it can pause running goroutines automatically without requiring to call `Gosched` directly.
Updates #3966
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Call runtime.Gosched() only when there is a work to steal from other workers.
Simplify the timeseriesWorker() and unpackWroker() code a bit by inlining stealTimeseriesWork() and stealUnpackWork().
This should reduce CPU usage when processing queries on systems with big number of CPU cores.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3966
Previously the selected time series were split evenly among available CPU cores
for further processing - e.g unpacking the data and applying the given rollup
function to the unpacked data.
Some time series could be processed slower than others.
This could result in uneven work distribution among available CPU cores,
e.g. some CPU cores could complete their work sooner than others.
This could slow down query execution.
The new algorithm allows stealing time series to process from other CPU cores
when all the local work is done. This should reduce the maximum time
needed for query execution (aka tail latency).
The new algorithm should also scale better on systems with many CPU cores,
since every CPU processes locally assigned time series without inter-CPU communications.
The inter-CPU communications are used only when all the local work is finished
and the pending work from other CPUs needs to be stealed.
Unpack time series with less than 4M samples in the currently running goroutine.
Previously a new goroutine was being started for unpacking the samples.
This was requiring additional memory allocations.
Usually the number of blocks returned per each time series during queries is around 4.
So it is a good idea to pre-allocate 4 block references per time series
in order to reduce the number of memory allocations.
* {app/vmstorage,app/vmselect}: add API to get list of existing tenants
* {app/vmstorage,app/vmselect}: add API to get list of existing tenants
* app/vmselect: fix error message
* {app/vmstorage,app/vmselect}: fix error messages
* app/vmselect: change log level for error handling
* wip
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
