K8s improvements #17
K8s improvements #17
Conversation
helm/public-people-api/values.yaml
Outdated
| @@ -10,7 +10,9 @@ image: | |||
| pullPolicy: IfNotPresent | |||
| cache_duration: 10m | |||
| resources: | |||
| requests: | |||
| memory: 12Mi | |||
There was a problem hiding this comment.
We have seen spikes of just shy of 60MB. If we say we can move to a instance with at least 12Mi it means we potentially could loose a pod if it spiked as there might not be enough memory on that instance to accomodate that spike? This is a public facing API where any outage is client facing so it would probably be better to air on the side of caution and request enough to deal with a spike???? Other applications I would be less concerned about
There was a problem hiding this comment.
That's a good question. I am not aware how fast the app would be able to request for more memory? I though it would be able to handle such spikes. The biggest spike I see in the last 30 days is 30 MB, I would increase the initial request to this, though this would mean more memory than average needed, but this amounts are small either way.
Adjusted memory requests and limit.
Parallel GTG, timed healthcheck.
Helm: pod anti-affinity, memory requests and limit.