Closed angelhof closed 1 year ago
I don't 100% understand... does the _problematic
case not fail when it should?
If we want to know what's mounted as what, /proc/mount
may be the easiest way to do that.
I don't 100% understand... does the
_problematic
case not fail when it should?
The problematic test does not lead to a crash or issue by overlay (even though it sounds pretty bad to just create a file in temproot). At the moment it only serves the purpose of testing that try
exits early if something unpredictable has happened to the overlay directories.
If we want to know what's mounted as what,
/proc/mount
may be the easiest way to do that.
Great, I made a new issue to exercise this behavior in a test.
This PR adds tests and fixes to address #41. Currently, the sandbox_valid check is not adequately extensive and we need to make it stricter.
In particular, one check that feels relevant is to determine whether the directory is already used by another overlay (which is supposedly not allowed based on the overlayfs documentation). This actually comes up in practice (I paste a manual test below) but I can't figure out how to determine it ahead of time (e.g., the workdir contains a work directory if the overlayfs is mounted or not).
Erroneous test where overlay is in use: