Bacula-users

Re: [Bacula-users] tape error and bacula: ...block numbers differ

2009-02-07 07:37:09
Subject: Re: [Bacula-users] tape error and bacula: ...block numbers differ
From: Kern Sibbald <kern AT sibbald DOT com>
To: Ralf Gross <Ralf-Lists AT ralfgross DOT de>
Date: Sat, 7 Feb 2009 12:40:44 +0100
On Saturday 07 February 2009 12:26:12 Ralf Gross wrote:
> Kern Sibbald schrieb:
> > On Saturday 07 February 2009 11:33:58 Ralf Gross wrote:
> > > Kern Sibbald schrieb:
> > > > The problem described in the email below is probably an important
> > > > data loss problem due (most likely) to an I/O error, but more
> > > > importantly due to a misconfigured tape drive.  From the information
> > > > I see below, it appears to me that you have lost significant data. 
> > > > This is probably due to the fact that the HBA or the tape drive is
> > > > running in asynchronous mode (cached or buffered) while Bacula
> > > > *requires* you to run everything concerning tapes in synchronous mode
> > > > -- please see the Tape testing chapter for more information.
> > > >
> > > > This data loss due to asynchronous mode will probably also occur when
> > > > your tape fills.
> > >
> > > The verify job hasn't found any errors, so it seems that all data
> > > that was backed up could be read.
> >
> > Verify will not find missing data, unless you restored everything and
> > compared it to the original, so I am still a bit skeptical ...
>
> Ok, understood.
>
> > > I still think it would be a good thing if bacula notifies the
> > > user about this problem by mail and in the final job output.
> >
> > Once the message is sent by Bacula, you have full control over how to
> > send it. It is just a question of configuring your Bacula.
>
> I think I misunderstand you here. Do you say bacula should have sent an
> message about this problem? Because bacula didn't and that's my point ;)

*Everything* that prints in a Job report is a message.  You included it in 
your email, therefore Bacula did send a message, and all messages are routed 
where you tell it to.  

It is possible that "MailOnError" is a misnomer (I didn't write that 
particular code) and really means "MailOnFail".  If you can verify that, I 
would accept an RFC to have a MailOnError and a MailOnFail or some other way 
perhaps that adds even more flexibility.

>
> This is the messages resource:
>
> Messages {
>   Name = "Regular Backup"
>   mailcommand = "/usr/sbin/bsmtp -h localhost -f \"Bacula\" -s \"Bacula: %t
> %e of %c %l\" %r" operatorcommand = "/usr/sbin/bsmtp -h localhost -f
> \"Bacula\" -s \"Bacula: Intervention needed for %j\" %r" operator =
> root@localhost = mount
>   operator = xxxxxx  = mount
>   MailOnError = root@localhost = all, !skipped, !terminate
>   MailOnError = xxxxxx = all, !skipped, !terminate
>   console = all, !skipped, !saved
>   append = "/var/log/bacula/bacula.log" = all, !skipped
> }
>
> I get notifications about other errors, but not this one.

Thought there is always the possiblity of a bug, I suspect that you are 
confusing an Error with a Fail.  Bacula in addition to many other message 
classes has Warning, Error, and Fail, all generate messages, only Fail kills 
the job.

Kern

>
> > I have modified the current development version so that any difference of
> > block numbers greater than one causes the Job to fail rather than simply
> > report an error.
>
> Thanks, Ralf



------------------------------------------------------------------------------
Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM)
software. With Adobe AIR, Ajax developers can use existing skills and code to
build responsive, highly engaging applications that combine the power of local
resources and data with the reach of the web. Download the Adobe AIR SDK and
Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users