Cloud native operator's pods cannot start

[Peter_Lepsis]

Hello,

we tried to deploy cloud native full stack operator and faced the issue that none of the pods can be started. The only error message we found was: "exec /usr/local/bin/entrypoint: exec format error"

The opanshift servers cannot access internet so the images are pullod from a private gitlab repo.

Here are the pods and their status:

[10:24 AM] u067054@mkb.hu (External)

NAME                                  READY   STATUS                  RESTARTS        AGE
dynatrace-oneagent-csi-driver-7vtmh   0/4     Pending                 0               8m45s
dynatrace-oneagent-csi-driver-7zxml   0/4     Pending                 0               8m45s
dynatrace-oneagent-csi-driver-cbpc9   0/4     Pending                 0               8m45s
dynatrace-oneagent-csi-driver-fkwzd   0/4     Pending                 0               8m45s
dynatrace-oneagent-csi-driver-kl8vf   0/4     Init:CrashLoopBackOff   6 (2m43s ago)   8m45s
dynatrace-oneagent-csi-driver-mbbvc   0/4     Init:CrashLoopBackOff   6 (2m35s ago)   8m45s
dynatrace-oneagent-csi-driver-msj9w   0/4     Init:CrashLoopBackOff   6 (2m42s ago)   8m45s
dynatrace-oneagent-csi-driver-nfm4g   0/4     Init:CrashLoopBackOff   6 (2m48s ago)   8m45s
dynatrace-oneagent-csi-driver-pvrzt   0/4     Init:CrashLoopBackOff   6 (2m23s ago)   8m45s
dynatrace-oneagent-csi-driver-sz7pz   0/4     Pending                 0               8m45s
dynatrace-oneagent-csi-driver-ttwdk   0/4     Pending                 0               8m45s
dynatrace-oneagent-csi-driver-wtbfx   0/4     Init:CrashLoopBackOff   6 (2m23s ago)   8m45s
dynatrace-oneagent-csi-driver-xmfww   0/4     Init:CrashLoopBackOff   6 (2m43s ago)   8m45s
dynatrace-oneagent-csi-driver-znq8t   0/4     Init:CrashLoopBackOff   6 (2m31s ago)   8m45s
dynatrace-operator-9bf6858bc-n6xsm    0/1     CrashLoopBackOff        6 (2m49s ago)   9m2s
dynatrace-webhook-75987cc554-6rtjv    0/1     CrashLoopBackOff        6 (3m10s ago)   9m2s
dynatrace-webhook-75987cc554-t5d2s    0/1     CrashLoopBackOff        6 (3m13s ago)   9m2s

 Thanks

Regards

Peter

[gopher · Accepted solution]

Hi @Peter_Lepsis,   

Sounds like you have the incorrect architecture type for nodes that you have in the code modules / one agent image.
I've seen this a few times when pulling an image from a MAC if a standard 'docker pull' is done without specifying the architecture type (generally by using digest), it changes the default architecture type to what is relevant for a MAC. 

It's not well documented, but there are a couple of potential options that you can use.

First is to use the the image & digest to specify the archive type.  e.g below

The second option is to use the following docker cli option when pulling the image

'--platform=linux/amd64'

At the end of the day, re-pull & replace the image with the correct architecture type and you should be good. 

Enjoy.

[Peter_Lepsis]

Dear gopher,

yes this was our first idea also. We deployed the operator in openshift with manifest and we removed all the architecture types other than amd64 but still not working

Thanks

Kind regards

Peter

[Julius_Loman]

@Peter_Lepsis based on your pods status the issue you have is with the operator image. Please ensure it matches your Open shift CPU architecture. Aren't you running OpenShift on Power ?