On Wed, Jun 10, 2009 at 3:04 PM, (private) HKS<hks.private AT gmail DOT com>
wrote:
> On Wed, Jun 10, 2009 at 2:19 PM, Hayden
> Katzenellenbogen<hayden AT nextlevelinternet DOT com> wrote:
>> When I run the btape test suite it completes no problem. I only get this
>> bug when I run btape fill over multiple tapes. It happens when it
>> switches tapes.
>>
>> If the current tape in the drive is not the tape from slot 1. When it
>> loads the tape in slot 1 I get the error then it carries on no problem.
>> Once it switches to the tape in slot 2 I get the same fatal error but it
>> completes the fill test without issues. Also running unfill does not
>> show the data being incorrect. If the tape in the drive is the tape from
>> slot 1 I only receive the error when it switches to the tape in slot 2.
>>
>> Just before I put this project on hold I had made the decision to ignore
>> the error and push on to see if it was just a btape issue.
>>
>> Good to know that I am no longer alone.
>>
>> H
>
> If the SD encounters a fatal error while writing a backup to tape, it
> will mark that backup as failed and you will not be able to restore
> from it even though the data is there (up through the end of the
> tape).
>
> I have a similar issue that I've not devoted the time to tracking
> down, where Bacula logs the error:
>
> 08-Jun 16:11 bufc1-sd JobId 3871: Fatal error: Unable to write EOF.
> ERR=dev.c:1669 ioctl MTWEOF error on "124T-Drive" (/dev/nrst0).
> ERR=Input/output error.
> 08-Jun 16:12 bufc1-sd JobId 3871: Error: Error writing final EOF to
> tape. This Volume may not be readable.
> dev.c:1669 ioctl MTWEOF error on "124T-Drive" (/dev/nrst0).
> ERR=Input/output error.
> 08-Jun 16:12 bufc1-sd JobId 3871: Fatal error: spool.c:294 Fatal
> append error on device "124T-Drive" (/dev/nrst0): ERR=dev.c:1669 ioctl
> MTWEOF error on "124T-Drive" (/dev/nrst0). ERR=Input/output error.
>
> This only happens when EOFing a tape. This was reported by btape on a
> multi-tape fill test as well, but the data was read back without
> error. For now, I just rerun the backup(s) that got jacked (I only do
> tape backups once a week).
>
> -HKS
>
Oh, and this is Bacula 2.2.8 on OpenBSD 4.4. I've not yet tested it on 3.
-HKS
>
>
>
>>
>> -----Original Message-----
>> From: Lars Nordin [mailto:Lars_Nordin AT mcafee DOT com]
>> Sent: Wednesday, June 10, 2009 11:03 AM
>> To: Hayden Katzenellenbogen
>> Subject: Re: [Bacula-users] Btape fill test fatal error WEOF
>>
>> On Wednesday 10 June 2009 13:03:43 Hayden Katzenellenbogen wrote:
>>> Lars,
>>>
>>> First off I found the first similarity. I installed Ubuntu 8.04LTS and
>> I
>>> am having the same issue. I never had a working installation before so
>>> not ground for comparison.
>>>
>>> If you are not too far down the rabbit hole maybe you could try
>> another
>>> flavor of Linux before putting too much time into this issue. This
>> could
>>> be something as simple as Ubuntu being the issue.
>>
>> It seems that we are not alone and it is not just Ubuntu.
>>
>> I submitted a Mantis ticket for the btape bug but just for the fact that
>> btape
>> reports a fatal error but still says that the test passed at the end -
>> this is
>> a bug with btape.
>> I figured I would ask if there other parameters to change and try that
>> before
>> filing a bug for btape fatal error itself.
>> Well, some one has already added a note to my report and he is using
>> Oracle's
>> RHEL clone and a different tape drive.
>>
>> ----------------------------------------------------------------------
>> (0004283) jbroughton (reporter) - 2009-06-10 18:27
>> http://bugs.bacula.org/view.php?id=1309#c4283
>> ----------------------------------------------------------------------
>> Verified by doing the following:
>> A) rewind tape, weof, then ran btape fill and got:
>> 10-Jun 11:19 btape: Fatal Error at dev.c:1705 because:
>> dev.c:1704 Attempt to WEOF on non-appendable Volume
>> B) Just in case, I rewound the tape again, weof, then created a volume
>> on
>> the tape from bconsole, and ran the btape fill, and recieved the same
>> error
>> message.
>>
>> Running Bacula 3.0.1 on Oracle Unbreakable Linux 5.2. Tape is a DDS-4.
>>
>>
>>> Also I am not 100% positive yet if this WEOF is actually a problem. Or
>>> just an erroneous error message.
>>>
>>> H
>>
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Crystal Reports - New Free Runtime and 30 Day Trial
>> Check out the new simplified licensing option that enables unlimited
>> royalty-free distribution of the report engine for externally facing
>> server and web deployment.
>> http://p.sf.net/sfu/businessobjects
>> _______________________________________________
>> Bacula-users mailing list
>> Bacula-users AT lists.sourceforge DOT net
>> https://lists.sourceforge.net/lists/listinfo/bacula-users
>>
>
------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables unlimited
royalty-free distribution of the report engine for externally facing
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|
