[timescale-multinode]feat: Add possibility to specify internal ip range #311
Conversation
|
|
griseau
changed the title
| @@ -79,6 +79,10 @@ spec: | |||
| echo "*:*:*:postgres:${POSTGRES_PASSWORD_DATA_NODE}" > "${PGDATA}/../.pgpass" | |||
| chown postgres:postgres "${PGDATA}/../.pgpass" "${PGDATA}/postgresql_helm_customizations.conf" | |||
| chmod 0600 "${PGDATA}/../.pgpass" | |||
| {{- if .Values.internalIpRange }} | |||
| echo "Adding {{ .Values.internalIpRange }} in pg_hba.conf" | |||
| echo "host all all {{ .Values.internalIpRange }} trust" >> ${PGDATA}/pg_hba.conf | |||
There was a problem hiding this comment.
Won't that be run on pod restarts too?
There was a problem hiding this comment.
I did not think about this case, but you're right, coming from Kubernetes documentation :
"If the Pod restarts, or is restarted, all init containers must execute again." https://kubernetes.io/docs/concepts/workloads/pods/init-containers/
It's also the case for the command executed just above this one. How would you handle it ? Override the whole file instead of just appending this new line ?
There was a problem hiding this comment.
Or I can do the same as it's done here
There was a problem hiding this comment.
@paulfantom updated according to your comment :)
There was a problem hiding this comment.
I was thinking about it a bit more and even with grep this may go sideways.
Let's assume someone is doing an upgrade of the helm chart and changed internalIpRange value. In such a scenario grep would fail and echo would be run again, which in turn would result in pg_hba.conf having old and new configuration options.
For me it look like we have 2 options:
- Force full management of
pg_hba.confvia ConfigMap. (I am not really a fan of this) - Create line markers and modify only options between line markers.
A bit more explanation on option 2. On the first run of the helm chart, we would add 3 lines, something like:
# START of the section managed by helm chart
host all all {{ .Values.internalIpRange }} trust
# END of the section managed by helm chart
Later in all subsequent runs, we can use grep to look for such a section and edit only the content of this section. This way even if internalIpRange changed on upgrade, helm chart would be able to update it accordingly.
When running inside a kubernetes cluster, internal ip range could be different (for example could be 10.108.0.0/14 on GCP).
This PR allow you to set the internal IP range so that it's added inside the pg_hba.conf file during init.
Happy to take feedback :)