Images don't seem to be multiarch?

[gz]

I'm installing this in a cluster of mixed machines (arm64 and amd64). Howevever after the helm installation I have failing pods on all the arm64 machines (amd64 works fine)

$ kubectl get pods -n observe -o wide
NAME                      READY   STATUS    RESTARTS     AGE   IP           NODE   NOMINATED NODE   READINESS GATES
core-dump-handler-5l7fq   1/1     Running   0            11s   10.0.6.140   ci4    <none>           <none>
core-dump-handler-5v7h4   1/1     Running   0            11s   10.0.2.26    ci2    <none>           <none>
core-dump-handler-fpvjg   1/1     Running   0            11s   10.0.4.93    ci3    <none>           <none>
core-dump-handler-kxc87   1/1     Running   0            11s   10.0.0.169   ci1    <none>           <none>
core-dump-handler-llxbt   0/1     Error     1 (3s ago)   11s   10.0.5.115   ci6    <none>           <none>
core-dump-handler-mws4j   0/1     Error     1 (3s ago)   11s   10.0.3.97    ci7    <none>           <none>
core-dump-handler-r5j5q   0/1     Error     1 (3s ago)   11s   10.0.7.68    ci8    <none>           <none>
core-dump-handler-xh9lh   0/1     Error     1 (3s ago)   11s   10.0.1.144   ci5    <none>           <none>

Looking at log reveals it's likely loading the amd image on the arm machines:

$ kubectl logs core-dump-handler-llxbt -n observe 

exec /app/core-dump-agent: exec format error 

From a glance at the build it looks like this projet is building multi-arch images; so I'm wondering if I'm doing something wrong

for reference this is my helm install command:

helm upgrade --install core-dump-handler core-dump-handler/core-dump-handler --create-namespace --namespace observe \
      --set image.limit_mem=256Mi \
      --set daemonset.s3AccessKey=... --set daemonset.s3Secret=... \
      --set daemonset.s3BucketName=... --set daemonset.s3Region=...

[No9]

Hey @gz
I cant seem to remember doing anything special in the charts for multiarch so I would say this is likely a default from helm/kubernetes
Not sure what best practice is for multiarch clusters but happy to take PRs if there is something we can do on this end

[jmallatt-kiva]

MUSL images are mentioned in the README

An MUSL based ARM build is available in https://quay.io/repository/icdh/core-dump-handler-musl Simply change the repository tag in the values.yaml

However, there doesn't seem to be a 9.0.0 build there. Would it be possible to get that image built?

[swors]

@jmallatt-kiva How you fix this?

[swors]

@No9 Can you push a musl image for 9.0.0？