ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1

[imranrazakhan]

We have following environment

• Three-nodes cluster deployed using chart timescaledb-single - 0.10.0 in test environment
• Kubernetes 1.22.3
• Storage is configured using local persistent volume.

• Backup is not configuered, i set backup=false in chart

  Today i want to clean disk so i scale down timescaledb pods and clean the disk from all three nodes and then scale up pods, but i am getting following error , am i missing something? is there any way to start from blank disk again?

# kubectl -n dev logs -f timescaledb-0 -c timescaledb
2021-12-04 21:55:53,775 ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1
2021-12-04 21:55:53,775 ERROR: failed to bootstrap (without leader)
2021-12-04 21:56:04,206 ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1
2021-12-04 21:56:04,207 ERROR: failed to bootstrap (without leader)
2021-12-04 21:56:14,207 ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1
2021-12-04 21:56:14,207 ERROR: failed to bootstrap (without leader)
2021-12-04 21:56:24,207 ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1
2021-12-04 21:56:24,207 ERROR: failed to bootstrap (without leader)
2021-12-04 21:56:34,207 ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1

I logged into timescaledb pod and check patroni status, this is first instance and why its role is replica rather than master?

$ patronictl list
+ Cluster: yq (uninitialized) +---------+---------+----+-----------+
| Member        | Host        | Role    | State   | TL | Lag in MB |
+---------------+-------------+---------+---------+----+-----------+
| timescaledb-0 | 10.244.0.78 | Replica | stopped |    |   unknown |
+---------------+-------------+---------+---------+----+-----------+

[davidandreoletti]

@imranrazakhan Delete k8s services from this chart (load balancer, nodeip related ones depending on your config) from the previous helm deployment should resolve this issue, as it did for me.

[imranrazakhan]

@davidandreoletti Thanks for updates i will check this, Can we have more insight why we have to delete services? is it related to endpoint? i check ep yaml file but couldn't find any hint which stopping us to do clean start?

[jholm117]

Having the same issue. Deleting the resources from the previous helm deployment did not solve the issue for me.

[bleggett]

Same issue - and confirmed no resources left in cluster from previous install.

[jholm117]

Having the same issue. Deleting the resources from the previous helm deployment did not solve the issue for me.

I was able to get this working eventually, it's possible I missed cleaning up an endpoint or something.

[imranrazakhan]

@jholm117 @davidandreoletti we can fix issue by just deleting one ep (EndPoint) with name like clustername-config, where clustername is name provided during helm installation.

[veereshhalagegowda]

I am still seeing this issue after using different release name and deleting older endpoints. It just stops suddenly after sometime. Any different solutions would be greatly appreciated. Thanks.

[jleni]

Same here. It is happening in the latest release 0.27.4 It resolves automatically after a few minutes

[jprecuch]

Same issue here. Happens on latest 0.27.5 as well. Would be good to see this fixed finally

[jfaldanam]

Same issue here, moving the deployment to a new namespace solved it temporarilly for me

[JohnTzoumas]

Removing endpoints from a previous helm deployment solved it for me.

[ayeks]

@JohnTzoumas thanks a lot! I have the same issue with the latest version.

I test disaster recovery right now and killed all PVCs + PODs. The startup of the new timescale pod stops at:

timescaledb 2023-05-25 11:53:56,422 ERROR: Error creating replica using method pgbackrest: /etc/timescaledb/scripts/pgbackrest_restore.sh exited with code=1

When I delete the 4 endpoints the recovery runs through.

[MSandro]

I have the same issue. But in my case I have disabled the persistent storage, because in our dev environment we would like to clean the db by just restarting the container. I have also tried to set this to false: patroni.postgresql.pgbackrest.keep_data = false but no effect.