github-mirrors/VictoriaMetrics
1
0
Fork 0
mirror of https://github.com/VictoriaMetrics/VictoriaMetrics.git synced 2024-11-21 14:44:00 +00:00
Code Issues Projects Releases Packages Wiki Activity

app/vmselect/netstorage: substitute sorting packedTimeseries with the natural order of the fetched blocks

Browse source 
This should minimize the number of disk seeks when reading data from temporary file.
This commit is contained in:
Aliaksandr Valialkin 2020-04-26 16:25:35 +03:00
parent 6954d0edb7
commit fcf57f9883
1 changed files with 13 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

30
app/vmselect/netstorage/netstorage.go
View file

@ -503,6 +503,7 @@ func ProcessSearchQuery(sq *storage.SearchQuery, fetchData bool, deadline Deadli
tbf := getTmpBlocksFile()
m := make(map[string][]tmpBlockAddr)
var orderedMetricNames []string
blocksRead := 0
bb := tmpBufPool.Get()
defer tmpBufPool.Put(bb)
@ -519,7 +520,11 @@ func ProcessSearchQuery(sq *storage.SearchQuery, fetchData bool, deadline Deadli
return nil, fmt.Errorf("timeout exceeded while fetching data block #%d from storage: %s", blocksRead, deadline.String())
}
metricName := sr.MetricBlock.MetricName
m[string(metricName)] = append(m[string(metricName)], addr)
addrs := m[string(metricName)]
if len(addrs) == 0 {
orderedMetricNames = append(orderedMetricNames, string(metricName))
}
m[string(metricName)] = append(addrs, addr)
}
if err := sr.Error(); err != nil {
putTmpBlocksFile(tbf)
@ -531,27 +536,18 @@ func ProcessSearchQuery(sq *storage.SearchQuery, fetchData bool, deadline Deadli
}
var rss Results
rss.packedTimeseries = make([]packedTimeseries, len(m))
rss.tr = tr
rss.fetchData = fetchData
rss.deadline = deadline
rss.tbf = tbf
i := 0
for metricName, addrs := range m {
pts := &rss.packedTimeseries[i]
i++
pts.metricName = metricName
pts.addrs = addrs
pts := make([]packedTimeseries, len(orderedMetricNames))
for i, metricName := range orderedMetricNames {
pts[i] = packedTimeseries{
metricName: metricName,
addrs: m[metricName],
}
}
// Sort rss.packedTimeseries by the first addr offset in order
// to reduce the number of disk seeks during unpacking in RunParallel.
// In this case tmpBlocksFile must be read almost sequentially.
sort.Slice(rss.packedTimeseries, func(i, j int) bool {
pts := rss.packedTimeseries
return pts[i].addrs[0].offset < pts[j].addrs[0].offset
})
rss.packedTimeseries = pts
return &rss, nil
}