[OpenAFS] Error reading diskDataHandle vol header

[Derrick Brashear]

On Tue, Apr 24, 2012 at 3:55 AM, Staffan H=E4m=E4l=E4 <sh@ltu.se> wrote:
> We had a crash on an AFS server yesterday. Afterwards, I noticed that vos=
 listvol had these lines in the end:
> ---
> **** Could not attach volume 538592892 ****
> **** Could not attach volume 538592823 ****
>
> Total volumes onLine 1618 ; Total volumes offLine 2 ; Total busy 0
> ---
>
> I've not been able to find any information about those two volumes. The v=
ldb seems not to know anything about them:
> $ vos exa 538592892
> Could not fetch the entry for volume number 538592892 from VLDB
> $ vos exa 538592823
> Could not fetch the entry for volume number 538592823 from VLDB
>
> Looking at the server logs, I find the following:
> SalvageLog:
> ...
> 04/23/2012 17:30:16 SALVAGING VOLUME 538592823.
> 04/23/2012 17:30:16 totalInodes 4
> ...
> 04/23/2012 17:30:17 SALVAGING VOLUME 538592892.
> 04/23/2012 17:30:17 totalInodes 4
> 04/23/2012 17:30:17 SALVAGING OF PARTITION /vicepb COMPLETED
>
> And, even though the salvager had run just a little while before:
>
> FileLog:
> Mon Apr 23 18:05:40 2012 VAttachVolume: Error reading diskDataHandle vol =
header /vicepb/V0538592823.vol; error=3D101
> Mon Apr 23 18:05:40 2012 VAttachVolume: Error attaching volume /vicepb/V0=
538592823.vol; volume needs salvage; error=3D101
> Mon Apr 23 18:05:41 2012 VAttachVolume: Error reading diskDataHandle vol =
header /vicepb/V0538592892.vol; error=3D101
> Mon Apr 23 18:05:41 2012 VAttachVolume: Error attaching volume /vicepb/V0=
538592892.vol; volume needs salvage; error=3D101
> Mon Apr 23 18:05:41 2012 Partition /vicepb: attached 1634 volumes; 2 volu=
mes not attached
>
> VolserLog:
> Tue Apr 24 08:12:33 2012 VAttachVolume: Error reading diskDataHandle vol =
header /vicepb/V0538592823.vol; error=3D101
> Tue Apr 24 08:12:33 2012 VAttachVolume: Error attaching volume /vicepb/V0=
538592823.vol; volume needs salvage; error=3D101
> Tue Apr 24 08:12:33 2012 1 Volser: ListVolumes: Could not attach volume 5=
38592823 (V0538592823.vol) error=3D101
> Tue Apr 24 08:12:34 2012 VAttachVolume: Error reading diskDataHandle vol =
header /vicepb/V0538592892.vol; error=3D101
> Tue Apr 24 08:12:34 2012 VAttachVolume: Error attaching volume /vicepb/V0=
538592892.vol; volume needs salvage; error=3D101
> Tue Apr 24 08:12:34 2012 1 Volser: ListVolumes: Could not attach volume 5=
38592892 (V0538592892.vol) error=3D101
>
> What should I do with them?

all 101 (VSALVAGE) tells you is that it needs to be salvaged. there
are several possible errors which can trigger the
Error reading diskDataHandle, depending on the state of the volume
data handle file inside /vicepb/AFSIDat for the volume, but really it
boils down to
1) does the file exist?
2) is it the right length?
3) is it the correct version and magic?

The volinfo program can tell you the name of the file, and potentially
fix it, if salvaging again does not, but it may be worth simply
salvaging individually those
two volumes first with bos salvage's -showlog parameter and looking at
the resulting lot to see what it tells you.

> They might even have been there for a long time, without anyone noticing =
before now.
>
> The server is a Redhat RHEL 4, running openafs 1.4.12.

Certainly lots of stuff fixed since 1.4.12.

--=20
Derrick