How to tune memory when running image in OpenShift

[dtsao]

Container platform

OCP 4

Version

postgresql-15:1-10 from https://catalog.redhat.com/software/containers/rhel8/postgresql-15/63d29a05fd1c4f5552a305b3?architecture=amd64&tag=1-10&push_date=1684253325000

running via a deployment config on Openshift 4.10

OS version of the container image

RHEL 8

Bugzilla, Jira

No response

Description

The readme for v15 refers to the --memory parameter. Where do I set the --memory parameter in an Openshift deployment config (DC) so that the shared_buffers and effective_cache_size are set automatically?

My DC has resource limits configured under spec:template:spec as:

      containers:
        - resources:
            limits:
              cpu: '1'
              memory: 4Gi
            requests:
              cpu: 100m
              memory: 400Mi

but my shared_buffers is the default 128MB:

# show shared_buffers;
 shared_buffers
----------------
 128MB
(1 row)

Do you recommend that I instead manually set the shared_buffers and effective_cache_size parameters by editing the postgresql.conf file located at /var/lib/pgsql/data/userdata/postgresql.conf?

---

FYI - For anyone who comes across this post due to errors like this in their logs:

ERROR:  could not resize shared memory segment "/PostgreSQL.1237418266" to 33554432 bytes: No space left on device

You should check the space allocated to /dev/shm:

sh-4.4$ df -h
Filesystem                            Size  Used Avail Use% Mounted on
...
shm                                    64M  1.1M   63M   2% /dev/shm
...

I increased the /dev/shm space from 64M to 128M via the technique here: https://stackoverflow.com/questions/43373463/how-to-increase-shm-size-of-a-kubernetes-container-shm-size-equivalent-of-doc/63152020#63152020

by adding lines to the DC:

spec:
  template:
    spec:
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 128Mi
	   containers:
          volumeMounts:
            - name: dshm
              mountPath: /dev/shm

After the above DC change, the could not resize shared memory segment errors have stopped.

Reproducer

No response

[fila43]

Off the top of my head, It seems that you tried to set up a higher memory limit than available, so the server set the default one.