Plugin fails when pod with PVC moves to a node with a different CPU architecture

[isaacdorfman]

Describe the bug

Plugin fails when installed when pod is on one CPU architecture and recreated on another.

Error message:
logger=plugins.backend.start t=2024-11-26T11:08:26.069257688Z level=error msg="Could not start plugin backend" pluginId=yesoreyeram-infinity-datasource error="fork/exec /var/lib/grafana/plugins/yesoreyeram-infinity-datasource/gpx_infinity_linux_amd64: no such file or directory"

Note the gpx_infinity_linux_amd64 the plugin was installed when on an arm64 node so gpx_infinity_linux_amd64 doesn't exist.

Steps to reproduce

1. Run the pod with a PVC on an amd64 node.
2. Install the plugin
3. Delete the pod and create another one mounting the same PVC on arm64 node.

[yesoreyeram]

I don't think this is a bug with the plugin. Each architecture require different plugin binary. So mounting the same volume is not the correct way.

If you think still this is a bug, I would recommend creating a bug in grafana repo instead of plugin repo as this is not specific to plugin.