cancel
Showing results for 
Search instead for 
Did you mean: 

maximum partially filled media - another "why is this is"

schrammd
Level 5

hi all,

I see this topic comes up from time to time but I also see that there supposadly was in the works the ability to override the EOM in order to force spanning I guess (in order to fill partially full active media). We are running 7.5.0.7, Solaris master, single RHEL media server with 5 LTO5 drives, and the pool Offsite-1mo is set to 1 partially fill media yet on any given day there is always 3-4 partially full tapes. The reason this irks me is we vault daily and I am consuming at least 2x more media than should be.

I've checked the basics - all drive types are the same, no mixed retentions of anything using that pool, and only 1 media agent. Media sharing is not enabled. It doesn't matter what I set the limit to for this pool (and another BTW), it just goes and pulls in new tapes from scratch every day leaving us needing to vault partially filled tapes all the time.

What is the secret to force it to fill those tapes before pulling in new from scratch?

thanks!

1 ACCEPTED SOLUTION

Accepted Solutions

mph999
Level 6
Employee Accredited

If you have >1 Max Partially full tapes set, you can, if a tape spans, get extra tapes used.

This as per the TN was planned to be addressed in a future release.

The end result of this, was that Engineering considered the change, but the end result was they declined as they stated NBU was working as designed.  I'll  edit the TN.

 

However, this is not what you are seeing, as far as I am aware if max full media is 1, it should only ever use one at a time as the trigger for new media is EOM, and providing you started with no more than 1 full tape, each tape used will be full (apart from the last of course as it's unlikely you backup fills exactly X tapes).  The suspect causes would be backups of different retention / different tape density but you have checked this.

I don't think I'm missing anything obvious, but I've asked Marianne to take a peep.  Else. I guess we'll have to look in the MDS logs to see what it does when it selects new media.

View solution in original post

12 REPLIES 12

schrammd
Level 5

oh and the reference to fixing this "feature" is at the bottom of this page https://support.symantec.com/en_US/article.TECH90564.html

 

mph999
Level 6
Employee Accredited

If you have >1 Max Partially full tapes set, you can, if a tape spans, get extra tapes used.

This as per the TN was planned to be addressed in a future release.

The end result of this, was that Engineering considered the change, but the end result was they declined as they stated NBU was working as designed.  I'll  edit the TN.

 

However, this is not what you are seeing, as far as I am aware if max full media is 1, it should only ever use one at a time as the trigger for new media is EOM, and providing you started with no more than 1 full tape, each tape used will be full (apart from the last of course as it's unlikely you backup fills exactly X tapes).  The suspect causes would be backups of different retention / different tape density but you have checked this.

I don't think I'm missing anything obvious, but I've asked Marianne to take a peep.  Else. I guess we'll have to look in the MDS logs to see what it does when it selects new media.

mph999
Level 6
Employee Accredited

Marianne didn't think of anything I'd  missed, but did send me a link which I have yet to read which may or may not be relevant. 

What's bugging me is that this isn't a commonly reported issue, yet your still having an issue with something that should work ...

Has this ever worked for you, or have you always had the issue. 

 

Marianne
Level 6
Partner    VIP    Accredited Certified

I was thinking of a TN that got removed by Symantec:

How to cause NetBackup to use and fill older media before selecting new media just added
http://www.symantec.com/docs/TECH35407

Contents:

VERITAS NetBackup (tm) tries to load balance tape usage, favoring media with the fewest mounts assigned first. When a new tape is added with zero mounts assigned, that tape would be a prime candidate for use first. In some installations with smaller tape libraries, it may be desirous to first fill the older media prior to writing to the new unwritten tapes, the end result being, those older media can be filled and replaced, maximizing tape usage.

In the Media section of the GUI, sort the contents by selecting the Robots tab to display those media currently in the library first. Review the Mounts column for each media to find the tapes with the highest number of mounts.
Add the new media to the library, then perform a robotic inventory.
Then for each of the new tapes just added, run the following command from the master server:

/usr/openv/volmgr/bin/vmchange -n <number> -m <mediaID>

where <number> = a mount count higher than any currently within the tape library.

schrammd
Level 5

do you recommend I open a case on this then? I have checked and rechecked the retention settings and such, and just cannot understand why this is happening. What I am wondering is if there is some override going on when using basic DSUs, doing multiple copies to seperate pools, or possibly the multiplexing setting on the DSUs. Either way, this sucks for us.

thanks again

mph999
Level 6
Employee Accredited

You could open a case if you believe it's not working as designed.  I'll test it as soon as I get time as I was interested in what you explain it is doing.

schrammd
Level 5

I'm not sure what logs would help you with this. I don't know what MDS logs means so if there are any particulars I can attach let me know. To reiterate the settings:

media pools:

Offsite-1mo set to max 1 partial filled. Daily has 2 active partially filled tapes which get vaulted daily

Local set to max 1 partial filled. Has 4 active partially filled tapes which stay onsite. There is always 2 being written to daily, 2 are of seperate dates last written.

Alll fulls go to DSU first, then staging schedule makes 2 copies, 1 to the offsite1mo pool keeping same retention 30days, 2nd copy to local pool.

thanks

mph999
Level 6
Employee Accredited

MDS (Media and Device selection)  is a 'sub component' of EMM.  It has a separate 'oid' (log number if you like) than EMM,

111 = emm log

143 = MDS

... and just to confuse things, there is no separate log file for MDS, it logs into EMM ...

 

From your description, the pools written to :

Offsite-1mo

Daily

Local

... and of these Daily and Local currently have >1 partially full tape.

schrammd
Level 5

There are 2 pools I see having the issue that causes extra consumption:Offsite-1mo and Local.

There is a Diff pool which is set to "2" and it has 2 tapes getting updated daily(as it should) and an active tape last updated 4/27 which is weird but since this is not affecting use of tapes from Scratch, I don't care much about it.

It's really the Offsite-1mo pool that is causing there to be way more tapes used and vaulted (and subsequently expensed) than needs to be.

thanks

Genericus
Moderator
Moderator
   VIP   

I have seen this issue where there are multiple media servers writing to one pool. While media server 1 is writing, media server 2 will wait, sometimes due to tape unmount delay, media server 2 will start a new tape, and presto, you have more partially full tapes.

Once you get a partially full tape in the pool, you cannot get rid of it, except by vaulting or filling it. NetBackup will always seep a scratch tape upon filling the current tape, so it will never lower the amount of partially full tapes.

 

NetBackup 9.1.0.1 on Solaris 11, writing to Data Domain 9800 7.7.4.0
duplicating via SLP to LTO5 & LTO8 in SL8500 via ACSLS

Genericus
Moderator
Moderator
   VIP   

LOL, if you are daring, based on the link above, theoretically, you could freeze all you scratch tapes, after looking for unassigned media in the pool, then unassigned scratch tapes, it is supposed to use partially full tapes already assigned... Never tried that one.

 

NetBackup 9.1.0.1 on Solaris 11, writing to Data Domain 9800 7.7.4.0
duplicating via SLP to LTO5 & LTO8 in SL8500 via ACSLS

mph999
Level 6
Employee Accredited

Unfortunately, you can't freeze a scratch tape ...