Bacula-users

Re: [Bacula-users] VXA-320 Autoloader followup

2008-04-25 15:29:14
Subject: Re: [Bacula-users] VXA-320 Autoloader followup
From: Brian Debelius <bdebelius AT intelesyscorp DOT com>
To: Mark Nienberg <gmane AT tippingmar DOT com>
Date: Fri, 25 Apr 2008 15:21:35 -0400
wow, thanks for the tip

Mark Nienberg wrote:
> Anyone who followed the original thread "tape read errors" started 3/3/2008 
> may be 
> interested in this followup.  I did a bit of testing with a variation of the 
> mtx-changer script on my VXA-320 Autoloader and noted that the time between a 
> successful load of a tape into the drive and when the "mx status" command 
> returned 
> "ONLINE" varies tremendously.  It is sometimes about 30 sec.  It is often 
> about 3 
> min.  Sometimes it is as much as 17 min.
>
> I spoke to support at Exabyte to see if they could explain this.  It is 
> expected 
> behavior. In fact, he told me that 20-30 min would not be unexpected.  The 
> drive 
> firmware is paranoid about the integrity of the system area on the tape 
> (normally 
> written at the very beginning).  Whenever the drive passes over the system 
> area (it 
> happens when tapes are loaded) it takes the opportunity to do some 
> diagnostics.  If 
> it has trouble reading the data there it will try several more times.  
> Sometimes it 
> will decide that the area is wearing or unreliable and it will move the 
> system data 
> to a blank section of tape farther away from the beginning.  It may decide to 
> re-tension the tape.  When a tape is written to the end of tape, it will put 
> another 
> copy of the system data there.  All of this means that the drive can spend a 
> fair 
> amount of time convincing itself that all is well with the newly inserted 
> tape. 
> After it is convinced, it will return ONLINE in response to a status query.
>
> I guess this paranoia on the part of the firmware is a good thing as it 
> improves 
> reliability.  In order to get bacula to gracefully handle the long wait I set
>
> Maximum Changer Wait = 1800  #(30 minutes)
>
> in the device resource and I modified the mtx-changer script to query for the 
> status 
> at 10 sec intervals to a max of 28 minutes.
>
> Mark Nienberg
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
> Don't miss this year's exciting event. There's still time to save $100. 
> Use priority code J8TL2D2. 
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
> _______________________________________________
> Bacula-users mailing list
> Bacula-users AT lists.sourceforge DOT net
> https://lists.sourceforge.net/lists/listinfo/bacula-users
>   


-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

<Prev in Thread] Current Thread [Next in Thread>