Is a Jottacloud repo reliable? Seeking user input

[kapitainsky]

@akrabu

My work around is to do “restic copy || restic copy || restic copy || …” since Copy is resumable, and each time it fails, it will try again, until you get the full snapshot down locally. Then you can restore from the local snapshot, instead of the poor quality backend.

This is clever tip! Ta

Re storage have a look at iDrive E2 S3. I am testing it now - so far so good. Realistically priced S3 storage.

[MichaelEischer]

@akrabu

What I personally wouldn’t mind is a “–retries” switch, where you can determine just how tenacious Restic will be. This would possibly make my work-around for poor quality connections unnecessary.¹ Either that, or make restores resumable. Or ideally, both!

The problem is not the number of retries, but rather that a successful download of a pack file part yields corrupt data. That requires retries on a different level than are implemented so far.

[akrabu]

Ah, yes, good point! Something like --retries and --low-level-retries perhaps. Not sure about the naming convention, but being able to tune both would be nice for power users. Or… desperate users lol

[nicnab]

@akrabu

I feel like that would be out of scope for Restic. It’s a backup program, after all - not a connection diagnostic program.

Oh I thought it’s a bitrot detector - at least that is what it’s usually doing for me Kidding aside you’re most likely right. And I’ve got to say that I have yet to hit a single real problem with restic. I only use SFTP and REST though and that is rock solid, no questions asked.

[MichaelEischer]

@akrabu

Ah, yes, good point! Something like --retries and --low-level-retries perhaps.

I’m not a fan of having dozens of mysterious options whose effects are entirely obscure to the general user. From the discussions in the thread, it looks like a retry if the blob is corrupted should be enough. So far I don’t see any compelling need to add new options.

Btw, I’ve opened Retry blob loading if backend returns corrupted data · Issue #4440 · restic/restic · GitHub

[marky]

Do you have a script for your compare file test?
Do you use the check data argument to check say 1GB or do you check specific files.

[mdl]

I just select one random file from source, restore that one to a temp location and compare it with the source. Nothing special really.

[mdl]

To get back on the original topic if Jottacloud is reliable. Like I wrote, I occasionally had errors with Jottacloud. Every since my storage was more then 5TB I had errors every day.

I have added the option --bwlimit 1M:1M to rclone and ever since I have had no errors (knock on wood).

[akrabu]

To conclude this thread for me, I switched to iDrive e2 instead. I didn’t like how I had to let Jottacloud “settle” for several weeks before I could reliably restore data from a large backup. iDrive e2 has been rock solid, and is a bit cheaper than B2 right now.