[Borgbackup] recommended check option

[Thomas Waldmann]

> Just 11.6 TB of compressed and deduplicated data (source data is about 
> 17.2 TB, ratio ~68%). Disks (both, productive and backup disks) are 
> mounted via 10GB ethernet line. It became slower day by day,

I don't see a reason why borg should get (significantly) slower over 
time as long as it does not run out of resources (like RAM) - besides 
the startup and shutdown overhead (of course, loading and saving big 
repo index / chunks index and files cache takes more time than smaller 
ones).

> and at he 
> end if the time borg runs was too short (less than 3 hours continuously) 
> the amount of backed up data was even negative.

How can a data amount be negative?

>> Interrupted backup (connection issue, transport endpoint not connected):
>> just start borg create again and again in the same way. It won't
>> transmit the data again which it already has in the repo. For this, a
>> "borg check" is not needed.
> 
> But what about the lock and the error messages? Should both disappear 
> after some borg create runs?

If there is a lock in the repo, you'ld need to follow to "borg 
break-lock" docs. But borg usually avoids leaving back locks and also 
can detect and auto-remove stale locks in some circumstances (== if it 
can be sure).

Running a rather recent borg version might help a bit with this, esp. if 
borg gets killed via some signal and the repo is via ssh:.

>> I/O Error: fs or storage device malfunctioning, if you have access, run
>> a SMART long test on the disks and check the SMART status/logs 
>> (smartctl).
> 
> It is on a Synology box. I just started scrubbing ("Datenbereinigung" in 
> german) and a S.M.A.R.T. long test. Additionally, it offers an IronWolf 
> Health check. Usually, scrubbing runs every 6 months, S.M.A.R.T. and 
> IronWolf Health monthly. I'll quote if any errors are reported.

OK, hopefully that will find the root cause of the IOError.