debug help to indentify partial reads on readonly overlay

we have a setup with a fuse-overlayfs (v1.8.2) that is a readonly filesystem of the underlying filesystem. the underlying filesystem is lustre, and that has cache logic that is annoying for certain usecases. the used config is plain and simple `fuse-overlayfs overlay /readonlymount -o ro,lowerdir=/lustremount,threaded=8,dev,suid`

it fixes the issues we had with lustre, except for one case that i would like to debug, but i don't know how or where to start.

the case gives in more then 50% of the tries on the readonly mount an error do to a partial read of a file (eg as part of the case, some python code is read and executed from it and fails with "unexpected end of file").

we never have the issue when trying it from lustre directly, but ofcourse i'm not saying the issue must be in overlayfs.

any pointers on where to start looking or how to make useful traces to find the culprit are welcome.

Begin with the reported fuse-overlayfs v1.8.2 command and the workload that executes Python from the readonly mount; reproduce the partial-read failure and collect useful traces. Done means identifying whether the fault is in fuse-overlayfs, Lustre, or their interaction, with evidence.

Written by the indexing model from the issue text.