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

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

Le 05/06/2026 à 23:30, mannino a écrit :
> I understand. Thank you for your responses. I shall look into --hash, 
> it's very good that it digests on the fly. I shall also get RAM of one 
> or both our servers checked whenever possible.
> 
>  From your standpoint, are both of my archives broken in a similar way 
> (i.e. including the archive from my last message)?
> 
> As for my suggestion of storing slice counter in the header, you're not 
> considering it after all? That reasoning should stand regardless of my 
> particular problem.
> 
Don't take any offense from my last and short message.

The time scale for support (hours, days) is not the same as the one for 
feature addition (weeks, months, years).

I cannot thus add a feature to just understand what's wrong in a backup 
process, in particular when we reached a step during the investigations, 
where the most probable cause (even if it sounds improbable) seems to be 
outside dar/libdar, until this probable cause is invalidated or not by 
real test (by facts, not by assumptions).

On the other hand, adding a feature needs planning, consideration of 
already existing features, architecture review, then code review, before 
defining how it will be implemented... the comes the coding time.

Note also that I have limited time available for dar/libdar and spending 
it to help for support requests is less time available for new features.

Last, dar release process is long as it includes a lot of testing and 
adds some code reviews... for me this is the guaranty to not be 
overwhelmed by support requests and by user dissatisfaction (in 
particular of myself as being also a user of dar!) The testing passed 
include all the features you have been using here, I'm confident enough 
that seen the amplitude of the corruption you have, I (an the many other 
users) would have also met it if it was just a bug dar/libdar... but the 
we will see.

Now, in you context, I explain what would be the best next step to 
follow according to my understanding of dar code and of the context you 
have... that's all.

Cheers,
Denis