Exception during "serve restic" operation

[naffit (Jens)]

Hi

I'm using rclone to transfer data from restic to a GDrive backend, and it usually works without any hickups.

However, I was running a restic prune operation to fix a few errors inside the restic repository, which takes some time.. a really really long time since restic is reading ~3.5TB of data that is placed within that repository.

Without going into details, restics long processingtime is divided into the following phases:

Phase 1 where restic reads all data and checks the current index for its packs took just over 52 hours.
Phase 2 there restic removes any unused data in older snapshots took around 19 minutes
Phase 3 where restic rewrites packs to actually remove the data took 7 minutes
Phase 4 where restic rebuilds the updated index for its packs was running in its 22nd hour (around 40% complete) when suddenly the operation stopped and this was logged:

---

rclone: Exception 0xc0000006 0x0 0x167145c 0x452e95
rclone: PC=0x452e95
rclone:

and everything stops and restic just aborts.

This is using restic v0.9.5 and rclone 1.47 running on 64bit Windows.

I have no log apart from a few lines that restic logged, and I am hesitant to restart the operation since it will take another ~74 hours (at least) where I am unable to perform any other backups..

Due to the long runtime I am also worried about adding --log-file, --log-format, --log-level and -v since it will probably generate a vast amount of logging that is more or less useless.

Please advice how/what I can do to debug the issue or how I can record some resonable and useful amount of logging..

/Jens

[ncw (Nick Craig-Wood)]

If you run with --log-file (without -v) then at least you will catch the backtrace which will go a long way towards working out what the problem is.

It might be worth checking if rclone is using up too much memory while it is running (it shouldn't doing serve restic).

You might also want to try 1.48 which was released recently.

[naffit (Jens)]

Roger, thanks for that.

I'll see when I have a window for running such a long operation again.

[naffit (Jens)]

Okay, so I had another go at the restic operation and this time rclone did not crash, and everything went smoothly.

I hope it was a one time even..

[ncw (Nick Craig-Wood)]

If you can get it to do it again and have a backtrace then please make a new issue on github with it in. Rclone should never make backtraces!

[naffit (Jens)]

Of course - however, no backtrack was logged after I ran it on Windows via restic with these options:

serve restic --stdio --config rclone-restic-gdrive.conf --cache-db-path --cache-dir --drive-impersonate

The last run that was successful never did log anything after I added --log-file to the options above, but I hope that is enough for any future problems that needs debugging..