server
server copied to clipboard
nextcloud
trigger a rescan when trying to fopen a file that exists in cache but not on disk
When the file index gets out of sync, any attempt by client to open a deleted (but still in cache) file will result in errors until an admin manually triggers a scan to repair it.
Rather than letting clients loop in errors forever we can just detect this error condition and repair the file index ourselves.
Note that this has no effect on primary object store as there file_exists will check the cache.
I am facing this on a weekly basis with my dev instance where the data folder get reset frequently! Thanks for diving into this!
