Re: [Dar-support] Need help determining sliced archive corruption

[Denis C. <dar...@fr...>]

Le 03/06/2026 à 21:49, mannino a écrit :
> 
[...]
> 
> Meanwhile, you suggested we rule out disk errors, transfer errors and 
> wrong order of slices. But, I'm afraid, faulty RAM won't get us anywhere 
> either because, as I have mentioned in the first message, I have two 
> backups like that, created 1 month apart, from two different systems 
> (different hardware, locations, age, usage patterns, etc. - only backup 
> mechanism and Ubuntu/dar versions are the same). Both backups report

Even if it sounds improbable, the corruption by the RAM (of the system 
where dar runs) should be first invalidated. A test on one of the 
concerned nodes would be sufficient. So whenever you have the 
opportunity (production constraint, I can understand), a memtest86+ 
would worth running for a whole cycle (yes it may take a couple of 
hour...depending on how fast and how large is the RAM).

In the past, dar already revealed RAM weaknesses that did not avoiding 
the OS to run normally...  memtest86 brought the explanation.

See also that some users make even bigger backups than what you seems 
doing, and so far (here I'm crossing my fingers!!!) no one reported any 
such data corruption... while 2.7.15 is already five years old. IMHO, 
there is something weird in your environment.

I would also suggest in your backup process to leverage the --hash 
option of dar (--hash sha512 or --hash whirlpool at least), checking the 
hash right after each slice generation and before slice usage would let 
you completely eliminate storage and network as a cause of the problem, 
as the hash is performed in memory on the data just before it is written 
to disk.

Though, the --hash option will not be able to reveal any problem 
regarding RAM issue....

I'm completely out of any other plausible track than memory/RAM issue, 
really... I really need you to test the RAM of one of your host where 
dar is running, before going further in investigations with more 
improbable hypothesis: eliminating the probable causes before 
considering the improbable ones...

Cheers,
Denis