Sick Rage stalls during move to rclone mount (gsuite)

I have NZBGet download files, then process with nzbToSickBeard, and Sick Beard just needs to do the final move to my rclone mount (gsuite).

This seems to work fine, for a while.

After an as yet undetermined amount of time, Sick Rage will simply stall and become unresponsive while doing a move.

Whenever this happens, I look at the rclone mount stats heartbeat and I see a transfer is happening, but its speed is 0.

From the shell, I can still access the mount, write to new files (e.g. echo foo > /mnt/storage/foo), delete files (e.g. rm /mnt/storage/foo), etc.

But Sick Rage will not respond. Even attempting to kill -9 Sick Rage will not work. The only thing that does work is to fusermount -u /mnt/storage. This is very odd. It’s supposed to be impossible for kill -9 to fail?

Then Sick Rage will immediately become responsive again. Then I can re-mount, and manually re-trigger post-processing from NZBGet, and then the same file which just failed, will succeed.

Is Rclone mount just a bit unstable still for writes? Is it random that sometimes a write will get stuck? Or is it that after a certain period of time, the next write will get stuck?

It’s unstable for writes but it does work. I don’t think the mounts have the same quality retry logic yet. That being said, I increased my --retries to 40 and that had helped a lot.

In my testing once the failure happens and exceeds the default retry (3) then it hangs the thread. I created an issue for this issue.

I just ran rclone mount again with increased logging, and this occurs just before the speed starts its inevitable descent towards 0:

2017/04/11 13:12:46 DEBUG : pacer: Rate limited, sleeping for 1.423731422s (1 consecutive low level retries)
2017/04/11 13:12:46 DEBUG : pacer: low level retry 1/1 (error googleapi: Error 403: Rate Limit Exceeded, rateLimitExceeded)

It seems that once this happens, that particular move operation is unrecoverable (but others and the mount in general via the shell are OK), and the process that initiated the move (Sick Rage) cannot even be killed. I have to unmount and remount to recover.

Eventually NZBGet will mark the post processing as failed because Sick Rage never responded (or stopped responding), which at least lets me know there was a problem.

I’ll try upping the retries. Could you link to the issue you created?

Alternatively, it seems the thing people generally do or suggest is to use rclone move instead of rclone mount – but the reason I want to use rclone mount is to allow Sick Rage to delete old lower quality versions after post processing a new higher quality version.

Having Sick Rage point to a local directory will also make it unable to scan existing files.

You can solve that part by not including quality in the name

Thanks, I saw that tip in another thread. Not sure I want to go that route, though. I like having the quality in the name, and that is especially important if the Sick Rage DB is ever deleted and I need to import a library of existing files.