Jump to content

Clients randomly idle for extended periods holding up backup


Recommended Posts

Hi all,

 

Specs: G4 533DP/512MB RAM/OS10.4.6/Retrospect Backup 6.1.126/Qualstar TLS-4440 with AIT-2/ATTO Express PCI UL4S

 

I have been faced with this issue on and off for quite some time now but it hasn't been frequent enough to warrant too much concern. However, lately it is more frequently occurring and needs to be resolved.

 

What is happening is that when I have checked the log for the previous night's backup I have found one of the clients on our network was showing that it sat idle for a very long time (last night it was 6 hours) and backed up zero files even though it showed that there was a large volume of files required to be backed up (again, last night's client had over 20GB of files to be backed up). One weekend on a Full backup it sat on a client for 56hrs. There's no other message about the fault and there is plenty of tape space, as well as other tapes to continue with (tapes that have previously been named and used with this BackupSet).

 

This always causes the remaining Macs to not get backed up - which is a real danger!

 

What might be causing the client to sit idle for so long for apparently no reason?

 

One thought I have is that I have 4 weekly rotating backup sets. This consists of 8 BackupServer scripts; 4 for the Production Macs and 4 for the Servers. Each set of 4 are always active for the same time periods. All I do is recycle the script of the one I want to use for the week, put in the right tapes and let it take care of itself.

 

Is this an OK practice or should I always be going into the other scripts and deactivating them?

 

I ask this because it just may be possible that one of the other scripts may be interfering with the clients at the same time as the active script.... just a stab in the dark, anyway.

 

Oh, and I completely rebuilt the system in May of this year (went from 10.3.9 to 10.4) and I'm sure I was experiencing this anomaly prior to the rebuild. Retrospect was freshly installed - no preferences carried over if my memory serves me correctly.

 

Regards,

 

Brett

Link to comment
Share on other sites

>one of the clients on our network

>it sat on a client

>the client

 

- Is this the same client machine every time?

 

> it just may be possible that one of the other scripts may be interfering

>with the clients at the same time as the active script

 

No, Retrospect won't allow a pending script to start until the current script is complete; that's why it's OK to schedule scripts just a minute apart, and know that they'll run consecutively.

 

- What are your settings/preferences in regards to client connections (both application wide and for the script)?

- What finally happened after the 6 hours passed in your example?

 

Dave

Link to comment
Share on other sites

Quote:

No, Retrospect won't allow a pending script to start until the current script is complete; that's why it's OK to schedule scripts just a minute apart, and know that they'll run consecutively.

 


Dave,

I agree with the first part of your sentence, and the second part is technically correct but misleading.

 

It's my experience that Retrospect has an odd scheduler, and doesn't run scripts in the scheduled order they became runnable. Instead, it seems to select from all presently runnable (past scheduled start) scripts in alphabetic order, regardless of the relative start times. So the only way that I have been able to get predictable order in our scripts (necessary because we walk the network first, shutting down all clients, then back up the quiet server in a series of server scripts that have to be done in a certain order, too) is to name all scripts alphabetically according to desired order of execution, schedule the first one at the given start time and then schedule all others at the same time, one minute later. Caused a lot of head scratching until I figured it out because, on incremental days, everything could be caused to execute in the desired order by staggering start times because execution times were short. But on "new media" days, when backups were long, multiple scripts would become runnable due to delays and not execute in the scheduled order (unless named alphabetically). Very odd.

 

This is the observed behavior ever since Retrospect 2.x. I've never seen it documented, but that's the way it seems to work.

 

Russ

Link to comment
Share on other sites

hi russ,

 

Quote:

This is the observed behavior ever since Retrospect 2.x. I've never seen it documented, but that's the way it seems to work.

 

 


 

i'll confirm. this is the way the current (and all previous) versions of Retrospect for Macintosh works. regretable as it may be.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...