Is this a right approach Or will it lead to problems?
I have done that and no problems ,
I first stopped Influxdb then emptied those directories
Thanks ! I need to restore the portable backup of whole influxdb. If any of the database exists, it does not restore the backup with msg "database may exist ". Hence to force the backup I have written script to first delete the datastore directories and bring the influxd up again then proceed for restore. It prints some errors about _internal database base but it is successfully restoring the backup for all databases. Do I need to change the approach ?
You mean to avoid the errors for _internal ? What are the errors ?
Here are the errors:
2019/06/07 15:28:20 Restoring shard 30 live from backup 20190605T172017Z.s30.tar.gz
2019/06/07 15:28:20 Meta info not found for shard 31 on database _internal. Skipping shard file 20190605T172017Z.s31.tar.gz
2019/06/07 15:28:20 Meta info not found for shard 32 on database _internal. Skipping shard file 20190605T172017Z.s32.tar.gz
2019/06/07 15:28:20 Meta info not found for shard 33 on database _internal. Skipping shard file 20190605T172017Z.s33.tar.gz
2019/06/07 15:28:20 Restoring shard 22 live from backup 20190605T172017Z.s22.tar.gz
Can this create issue ?
It would be interesting to know why these errors occur , but it will be no issue .
Your database can function without _interal , you can even disable the _internal statistics gathering in the configuration file if you don’t want them.
@MarcV Oh! then we do not really care about internal stat gathering. Please confirm that _internal does not contain any other functional data hence we will not face any partial backup issues like missing retention policies or continuous queries or measurements or something similar. We don’t care as far as our whole databases will be restored completely.
I would like to file a bug for this case. can you please guide me in that as well?