You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In my relatively fresh OpenShift 4.4 cluster, there are 935 secrets.
The TokenController's watcher is pulling all of them into a cache, which inflates memory usage unnecessarily from 15 MB to 103 M (it eventually decreases to 44M).
I see 2 options:
Inject a custom cache that's always a "miss" for reads.
Fixes OLM metadata so it will install properly from Operator Hub for v0.2+. Adds make tasks to assists OLM testing.
Bumps the memory limit to mitigate an OOMKilled issue when running in OpenShift 4.4 (i.e. lots of built-in secrets). Opened #199 to look into it further.
Also update TokenController to skip processing any secret not using the well-known names. e.g. `secret-ibm-cloud-operator` `<namespace>-secret-ibm-cloud-operator`
Fixes OLM metadata so it will install properly from Operator Hub for v0.2+. Adds make tasks to assists OLM testing.
Bumps the memory limit to mitigate an OOMKilled issue when running in OpenShift 4.4 (i.e. lots of built-in secrets). Opened #199 to look into it further.
Also update TokenController to skip processing any secret not using the well-known names. e.g. `secret-ibm-cloud-operator` `<namespace>-secret-ibm-cloud-operator`
Signed-off-by: Art Berger <[email protected]>
In my relatively fresh OpenShift 4.4 cluster, there are 935 secrets.
The
TokenController
's watcher is pulling all of them into a cache, which inflates memory usage unnecessarily from15 MB
to103 M
(it eventually decreases to44M
).I see 2 options:
Since this is an order of magnitude difference, I think we'll need to opt for a no-op cache in the interim. An example using controller-runtime v0.6:
The text was updated successfully, but these errors were encountered: