ae3cb8e
@@ -137,6 +137,11 @@
imagePullPolicy: Always
ports:
- containerPort: 5432
+ protocol: TCP
+ resources:
+ limits:
+ memory: 512Mi
+ cpu: 0.4
readinessProbe:
timeoutSeconds: 1
initialDelaySeconds: 5
@@ -259,6 +264,7 @@
resources:
limits:
memory: 384Mi
+ cpu: 0.3
volumes:
- name: config-volume
secret:
@@ -345,6 +351,7 @@
Otherwise PostgreSQL will run out of memory in projects where the default limits are lower than this.
Signed-off-by: Hunor Csomortáni csomh@redhat.com
@mprahl or @ralph: please take a look.
@csomh looks good to me. Any reason you picked 0.4 instead of a nice round 0.5 CPU limit? :)
0.4
0.5
rebased onto a05a35ee1cfe07affc3aa53736b5272b3d18b893
rebased onto a718f9ab3baf54674b0239e09c946925001e2c00
That's a good question @mprahl! I have no clue :p
Right now I would guess that it came from the default CPU limit for UpShift projects (300 millicores), increased with 100 millicores, which made things 'feel' better.
I've quickly read through these docs again, but could not find anything about recommended strategies for setting these limits.
FYI: I've noticed that the resultsdb DeployementConfigs were missing CPU limits, so I've added those too. Better to be explicit about these things, they say :)
Okay, still :thumbsup:
:+1:
If this is fine by you, I'll happily merge this after it's rebased to current HEAD. Let me know, from the comments, this seemed to be under development/discussion still :) Thanks!
@csomh is away for a few days. Let's wait for him to come back to see what he thinks.
rebased onto ae3cb8e
This was rebased, it's ready to merge, @jskladan. Thanks!
Pull-Request has been merged by jskladan
Otherwise PostgreSQL will run out of memory in projects where
the default limits are lower than this.
Signed-off-by: Hunor Csomortáni csomh@redhat.com