scripted backups stopped running on G5 OSX Server

[krel]

My clients system is: PowerMac G5 Dual 2 Ghz Server. (2) internal 150 Gb drives striped as a 300 Gb array. (2) external firewire 800 170 Gb drives striped as a 340 Gb array. The Server was running OS X 10.3.9 and Retrospect 6.1.126 and driver update 6.1.5.102.

 

I updated Retrospect on my clients server about two weeks ago. I planned on upgrading their server to 10.4 today but things didn't go well. That's an under statement, things went very badly. I popped in the new system disk and rebooted. About the time I rebooted I realized I needed to make sure the backup had run that morning before continuing. As soon as G5 booted, I rebooted again to let it come back up on it's internal drive. It never rebooted on the internal RAID. I got the missing system icon so I rebooted on the system install disk thinking it had just lost it's boot volume preference. Once booted no drives were available to boot on in the startup disk control panel. I tried disk utilities which reported that one of my drives in the RAID had failed. No problem I think, It's early in the day, I'll just restore from backup. I quickly dropped a system on the backup volume and rebooted. Once rebooted I opened Retrospect to discover that the last update to the backup catalog was a week ago, May 1st. I did everything imaginable to resurrect the dead volume but it was hopeless. I reinitialized and restored from backup. Once I restored the boot volume I rebooted on it to check Retrospect logs. Retrospect had logged that it started a backup the morning of the 1st but there were no other later log entries. The log showed it had been running correctly for a week since I updated Retrospect. If Retrospect unexpectedly quit shouldn't it have launched and tried to run again on it's next scheduled event? Can I check more detailed logs somewhere to try to find out what caused the failure?

 

I use Retrospect at several client locations and I've never had this happen. I need to know why before I can trust it again. This particular clients office is 60 miles from mine and I don't visit them more than once a week. I can't loose a weeks backup because Retrospect's automated backup stops running. This week was very bad because between the backup failing on the 1st and the drive failing on the 8th they converted their accounting system and did payroll, all stored on the server. My client tells me this took them three days and that it cost him 10,000 dollars in wages and consulting fees. He wants to know why his data wasn't backed up since I told him it was set up to back up every morning before his business opened. His current opinion is that Macs and Retrospect sux. I would like to be able to convince him that neither is true.

 

In my own defense I originally setup this server for speed. It was used to store huge graphics files between prep and production. That is why it was striped and not mirrored. Had I intended it to be an accounting server it would have been mirrored. I knew the client intended to start using it for accounting files and I was planning on a complete reconfigure this week. Apparently I was a week late.

[rhwalker]

There's really not enough info in your post to figure out what happened. Log info and configuration info would be useful. That aside, and completely unrelated, you don't say what type of RAID is in use. Is this the (known to have problems) Apple software RAID on 10.3.9 or SoftRAID (known OK) or some Hardware RAID card (who knows) ? Some of the RAID solutions never report drive failure until manually interrogated, and, even then, may not allow easy recovery.

 

 

 

Doesn't your client (or you) get the daily emailed reports from Retrospect that report success/failure of the backup? Or didn't you set that up? That is a simple way to get assurance of the backup state.

 

 

 

Not to say that Retrospect has been perfect, but I have never seen your particular concern (Retrospect stops backing up, undetected) in over 10 years of use on ASIP servers and in the past year on our Xserve G5. But it sounds like you have more fundamental problems here than Retrospect, many of which require better policy and strategy on the front end. For example, an undetected broken tape or failed backup hardware could cause exactly the situation you describe, and proper monitoring (e.g., setting up the daily email log reports from Retrospect) would quickly catch such an issue.

 

 

 

Heck, if your only concern is the 60 mile drive to the client, why not VPN in to the client and use ARD to check Retrospect occasionally (I've done that from across the country to our server while on vacation, works fine), or get the daily backup reports emailed to you (and to your client, if the client wants them too) from Retrospect? email techology is a solved problem these days.

 

 

 

Russ