Segmentation Fault while sync (SnapRaid 5.3)

[ortega000]

I have this problem when I try to sync.

My system is on Debian Squeeze (OpenMediaVault).
2 x 2.0Tb drives (parity and qparity)
4 x 2.0Tb drives (data)
2 x 1.5Tb drives (data)
1 x 1.0Tb drive (data)

CPU Celeron 1.9Ghz
RAM 4Gb

After several hours the snapraid process dies, and I get this error.

There is something I can try for "debug" this problem ?
Thank you in advance.

Ciao,
ho questo problema quando provo ad effettuare il "sync" dell'array.
per ora ho provato ad effettuare il sync con 2 dischi di parita' (ancora non ho provato con un solo disco).
Ho fatto un test della memoria ram (con memtest86) e non risulta nessun errore.
C'e' qualcos'altro da testare per capire il motivo del "Segmentation Fault" ?
Grazie sin d'ora per ogni eventuale suggerimento.

[Andrea Mazzoleni]

Hi ortega000,

To analyze the problem you can use the "gdb" debugger. I'm really interested in the result.

Run:

gdb snapraid
r <same options and command you use for snapraid, like "sync"]
<now here it should segfault>
bt full
<please report whatever is printed into the screen>

For example:

gdb snapraid
r sync
bt full

Thanks,
Andrea

[ortega000]

Thank you Andrea for your quick answer, I will definitely try the "gdb" debugger and will post the results.
Right now I'm not able to test it because my system in syncing the array (I give it another try after using the command "shred"+format on both parity drives).
The ETA is about 18 hours so I think that I can start the test tomorrow evening.
Thank you again for your help.

[Andrea Mazzoleni]

Hi,

Another thing to try, is to recompile with disabled optimization. This will improve the reported info by gdb:

make distclean
./configure --enable-debug
make
gdb ./snapraid
r sync <or another command/option>
bt full
quit

Note that in this case we use "./snapraid" instead of "snapraid" because we want to run the one just compiled in the current directory.

In case "bt full" reports too much information to scroll back, just use "bt" for a shorter version.

Ciao,
Andrea

[ortega000]

After almost 24 hours the "snapraid sync" ended successfully, without the "Segmentation Fault".
The only difference is the "shred" command and the ext4 formatting for the two parity drives.
After the last sync I have used the gbd procedure you suggested, but there wasn't errors of any kind.
One last question: is possible that the "Segmentation Fault" was generated by a cronjob backup script executed during the sync (the script copied/moved files to different directories) ?
I have discovered that this script was executed during the sync that generated the Segmentation Fault error.

Ciao e grazie

[Andrea Mazzoleni]

Hi ortega000,

Yes. It's possible that an error detection triggers an execution path that contains a bug. In fact, the 5.0 version had exactly this kind of bug, exactly when a file is modified during a sync.

Are you sure that the problem didn't happened using 5.0 ? Anyway, I checked 5.3 and I wasn't unable to find a possible problem. But I'll do some more tests.

Ciao,
Andrea

[ortega000]

Thank you Andrea for your answer, sorry for the delay but I was out of town and without my laptop.
The version I used was the 5.3 (I used the OpenMediaVault plugin). Now I have upgraded the plugin with the latest SnapRaid (version 6.0).
I have used SnapRaid again but this time without errors.
Maybe is fault of my hardware (mobo, sata controllers, nic) I should run a stability test of my server PC. Thanks again for your support and for the SnapRaid program, I really like it :)
Bye