We are observing that many files are left open for ages, maybe forever, either in read or write mode. Probably the (fuse) clients died or disappeared without issue a correct close, but there seems to be no timeout and it stays this way.
That makes the reported “ropen” and “wopen” values inaccurate, but this is not the main problem.
Some files are kept open for writing, and have incorrect checksum or size in MGM. Issuing eos file verify /path/to/file -checksum -commitchecksum -commitsize is ignored, because FST says file is currently opened for writing id=1d8d71c2 on fs=689 - skipping verification (once per minute), so they can’t be fixed.
Shutting down and restart the FST for sure does reset the open state, but is there some other less destructive way to do this ?