short write error keeps appearing in 1.8.5 while everything goes well in 1.6.6

[chengshiwen]

Steps to reproduce:
List the minimal actions needed to reproduce the behavior.

I don't know how to reproduce. If a measurement is dropped and a large amount of data of the measurement is written suddenly, it is possible

Expected behavior:
Describe what you expected to happen.
Bug fixed

Actual behavior:
Describe What actually happened.

Environment info:

• System info: Run uname -srm and copy the output here
• InfluxDB version: Run influxd version and copy the output here
• Other relevant environment details: Container runtime, disk info, etc

InfluxDB 1.8.5, Docker

If I downgrade to 1.6.6, everything goes well

Config:
Copy any non-default config values here or attach the full config as a gist or file.

INFLUXDB_DATA_INDEX_VERSION=tsi1
INFLUXDB_DATA_WAL_FSYNC_DELAY=10ms

[lifeibo]

It maybe a bug while DropMeasurement.

--- a/tsdb/index/tsi1/log_file.go
+++ b/tsdb/index/tsi1/log_file.go
@@ -1121,8 +1121,6 @@ func (f *LogFile) seriesSketches() (sketch, tSketch estimator.Sketch, err error)
 }

 func (f *LogFile) ExecEntries(entries []LogEntry) error {
-       f.mu.Lock()
-       defer f.mu.Unlock()
        for i := range entries {
                entry := &entries[i]
                if err := f.appendEntry(entry); err != nil {
@@ -1134,6 +1132,9 @@ func (f *LogFile) ExecEntries(entries []LogEntry) error {
 }

 func (f *LogFile) Writes(entries []LogEntry) error {
+       f.mu.Lock()
+       defer f.mu.Unlock()
+
        if err := f.ExecEntries(entries); err != nil {
                return err
        }

	// Flush buffer and sync to disk.
	return f.FlushAndSync()

[chengshiwen]

@lifeibo Yes! I did execute drop measurement first!

[altery]

We also experience these short write errors from time to time. Also, we had deletion of measurements shortly before the errors started to appear.
We have to restart influxdb to recover from that situation.

[altery]

Is it known, if influxdb 2.x is also affected by this problem?