app/vmselect: add -search.cacheTimestampOffset command-line flag

This flag can be used for removing gaps on graphs if the difference between the current time
and the timestamps from the ingested data exceeds 5 minutes.

This is the case when the time between data sources and VictoriaMetrics is improperly synchronized.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/312
This commit is contained in:
Aliaksandr Valialkin 2020-02-21 13:54:10 +02:00
parent 539139391c
commit ec9bf39b5b
2 changed files with 15 additions and 3 deletions

View File

@ -848,6 +848,10 @@ The most interesting metrics are:
of data loss stored in the broken parts. In the future, `vmrecover` tool will be created of data loss stored in the broken parts. In the future, `vmrecover` tool will be created
for automatic recovering from such errors. 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.
### Backfilling ### Backfilling
@ -858,6 +862,9 @@ It is recommended disabling query cache with `-search.disableCache` command-line
historical data with timestamps from the past, since the cache assumes that the data is written with historical data with timestamps from the past, since the cache assumes that the data is written with
the current timestamps. Query cache can be enabled after the backfilling is complete. the current timestamps. Query cache can be enabled after the backfilling is complete.
An alternative solution is to query `/internal/resetRollupResultCache` url after backfilling is complete. This will reset
the query cache, which could contain incomplete data cached during the backfilling.
### Profiling ### Profiling

View File

@ -18,7 +18,12 @@ import (
"github.com/VictoriaMetrics/metrics" "github.com/VictoriaMetrics/metrics"
) )
var disableCache = flag.Bool("search.disableCache", false, "Whether to disable response caching. This may be useful during data backfilling") var (
disableCache = flag.Bool("search.disableCache", false, "Whether to disable response caching. This may be useful during data backfilling")
cacheTimestampOffset = flag.Duration("search.cacheTimestampOffset", 5*time.Minute, "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")
)
var rollupResultCacheV = &rollupResultCache{ var rollupResultCacheV = &rollupResultCache{
c: workingsetcache.New(1024*1024, time.Hour), // This is a cache for testing. c: workingsetcache.New(1024*1024, time.Hour), // This is a cache for testing.
@ -215,10 +220,10 @@ func (rrc *rollupResultCache) Put(ec *EvalConfig, expr metricsql.Expr, window in
return return
} }
// Remove values up to currentTime - step - maxSilenceInterval, // Remove values up to currentTime - step - cacheTimestampOffset,
// since these values may be added later. // since these values may be added later.
timestamps := tss[0].Timestamps timestamps := tss[0].Timestamps
deadline := (time.Now().UnixNano() / 1e6) - ec.Step - maxSilenceInterval deadline := (time.Now().UnixNano() / 1e6) - ec.Step - cacheTimestampOffset.Milliseconds()
i := len(timestamps) - 1 i := len(timestamps) - 1
for i >= 0 && timestamps[i] > deadline { for i >= 0 && timestamps[i] > deadline {
i-- i--