Re: [Dar-support] Exception in cache.cpp
For full, incremental, compressed and encrypted backups or archives
Brought to you by:
edrusb
|
From: Denis C. <dar...@fr...> - 2021-12-30 21:50:09
|
On 27/12/2021 15:20, Robin Atwood wrote: > On Mon, 27 Dec 2021 10:24:19 +0100 > Matus UHLAR - fantomas <uh...@fa...> wrote: > >>>> On 26.12.21 14:57, Robin Atwood wrote: >>>>> While performing a full backup last night I suddenly got this >>>>> exception: >>>>> >>>>> Final memory cleanup... >>>>> ---- exception type = [BUG] ---------- >>>>> [source] >>>>> File cache.cpp line 562 : it seems to be a bug here >>>>> It is quite hard, repeating the run produced exactly the same >>>>> result.I have a completely current Gentoo system; dar is version >>>>> 2.7.2 and was updated on 08-10-2021; glibc is 2.34-r4. >> >>> On Sun, 26 Dec 2021 17:45:10 +0100 >>> Matus UHLAR - fantomas <uh...@fa...> wrote: >>>> I guess compiler version is important too. >> >> On 27.12.21 15:54, Robin Atwood wrote: >>> Gcc is at 11.2.1. But dar has been running in its current >>> configuration for over two months; >> >> havent't you changed anything on the gentoo system? >> change of a library can cause problems similar to this one. >> (and it's often quite hard to track because of gentoo nature). >> >>> I wonder if it is an environment problem. The >>> exception doesn't occur when I do a differential backup so I guess >>> archive size plays a part. >> >> Hi, > > OK, I have fixed the problem. I noticed an I/O error on the USB drive I > backup to so I ran fsck on the partition and that allowed the backup to > complete. It would be nice if dar could issue a helpful error message > when it gets an I/O error but maybe that is not easy to do. OK good to know the reason that lead dar to report this message. I will try to reproduce it and see how this message showed up. Dar normally propagate the system errors (including I/O errors) to the user as is, but here it seems the filesystem turned to read-only during the backup process and this lead dar to consider the problem to be an internal error rather than an system error (where from this type of error message). > > Thanks > Robin > Cheers, Denis |