You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Whilst investigating #10052 I triggered another bug, which I initially thought was the source of #10052, but in the end isn't. The bug manifests itself as follows:
You start up the sever with some data in the WAL.
The data is added to the cache on startup.
The cache snapshots the data to a TSM file (the cache is empty).
You delete the data, e.g., via DROP SERIES or DROP MEASUREMENT.
The data is deleted from the TSM files via tombstone markers.
You restart the server - the data is back again.
You can reproduce via:
- startup influxdb
> insert cpu c=12
> select * from cpu
name: cpu
time c
---- -
2019-06-28T12:04:11.671819Z 12
- Shut down server
- Start it up, but with a much lower snapshot time so you don't need to wait..
$ INFLUXDB_DATA_CACHE_SNAPSHOT_WRITE_COLD_DURATION=10s influxd
- *important*, wait for the cache to snapshot to a TSM file.
> drop measurement cpu
> select * from cpu
>
- restart server
> select * from cpu
name: cpu
time c
---- -
2019-06-28T12:04:11.671819Z 12
This affects all versions back to at least 1.5.3. I didn't test on earlier versions.
The text was updated successfully, but these errors were encountered:
Whilst investigating #10052 I triggered another bug, which I initially thought was the source of #10052, but in the end isn't. The bug manifests itself as follows:
DROP SERIES
orDROP MEASUREMENT
.You can reproduce via:
This affects all versions back to at least
1.5.3
. I didn't test on earlier versions.The text was updated successfully, but these errors were encountered: