Low storage failure to prune

[bquandt]

Low storage failure to prune

[x42]

The prune REST call returns Error dispatch server is not running.

This is because (since the disk-is full) the server can not update its alive marker in the database and the REST call does not contact the server.

Proposed solution: the prune REST call does not check if the server is alive (timestamp in database) but simply if there's a PID for the server in the database. -- downside: the call may fail silently.

[x42]

sorry wrong button pressed - this is not fixed yet.

[x42]

partly fixed in git rev 8787a7f (check for PID only). That still fails on reboot or server re-start: the server won't be able to set its new PID in the database)

question: how did you end up with a disk-full in the first place? With default config will prevent new recordings (or delete old ones) if the disk is close to full. It should take quite some config-file hacks to get into this state.

OTOH it may happen during postprocessing (concatenating files) and/or log-files.

[bquandt]

Didn't do anything special. Just let the thing run until no space. I didn't do anything special to change the config's, not extraoridinary, or beyond what would probably be the expected usages..

[x42]

fixed in ee62bbc279

• server makes space to save database when it is restarted (it deletes the oldest audio files if it must) to save its new PID and update the database.
• prune REST call ignores server uptime and checks for server PID only

What we really should do is prevent the disk full in the first place.