Improve s3gw telemetry extraInfo

[irq0]

What needs to be done

202 has the discussion.

Wish list:

• process memory consumption. e.g PSS collected from smaps awk '/Pss:/ { pss+=$2 } END { print pss }' /proc/$(pgrep radosgw)/smaps)
• object/bucket counts (plus logical size)
  • select count(*) as sfs_bucket_count from buckets;
  • select count(*) as sfs_object_count from objects;
  • select sum(size)/1024/1024 as sfs_object_size_total_mb from versioned_objects where object_state = / ObjectState::COMITTED /;
• k8s version
• PVC type
• pod_name, container_name, container_image, container_hostname, pod_namespace.

Why it needs to be done

See #202

Acceptance Criteria

Data shows up on the grafana dashboard

Additional Information

[l-mb]

If we get this in in the beginning we don't have to explain changing the telemetry reporting later via release notes and possibly get consent again :)

[m-ildefons]

Quite interesting information would probably a "deployment mechanism". This could be an environment variable that is injected into the Pod and set by the deployment mechanism (Helm/Longhorn) at deployment time. Then we get insight into how popular standalone vs. bundled s3gw is.