I started to test backup/restore tools for influxdb 1.5.4.
Basically, I followed the process as described in Remote restore not working - "DB metadata not changed. database may already exist" · Issue #9593 · influxdata/influxdb · GitHub :
- Back up database: influxd backup -portable /tmp/backup
- Restore to a new database: influxd restore –portable –db mon –newdb mon-restore
- copy data via select statement (CLI):
use mon-restore
**SELECT * INTO mon…:MEASUREMENT FROM /.*/ GROUP BY ***
Basically, this is working.
However, I’ve got a problem, regarding retention policies:
Since some time passes between the backup and the restore, same data in the backup is “too old”.
What happens is:
The restore seems to stop with the error message:
“ERR: partial write: points beyond retention policy dropped=”
This sounds reasonable, but in fact just “some” of the metrics data have been inserted.
It seems (I didn’t check in detail) that the restore stopped when trying to insert the data “too old”.
To test further, I extended the retention policy on the original database so that all data can be inserted. That worked, all data has been transferred now.
However, I can’t see how that really works for production databases: Changing the retention policy temporarily doesn’t seem to be a good solution.
My question is now:
Is this working as expected?
If yes: How to deal with this without changing temporarily retention policies?
My expectation would have been that data that doesn’t fit with the retention policy won’t be inserted, but the query will be executed until the end.
