"Closing" going on for 24 hours on Panther client; repeatable; really stuck

[scstr]

Hello,

 

I've hunted the forums for help. The only thing close to my problem was here.

 

http://forums.dantz.com/ubbthreads/showflat.php?Cat=&Number=33287&page=1&view=collapsed&sb=5&o=&fpart=1

 

Except in my case, I run Retrospect Server 5.1.167 on Mac OS 10.2.8 on a Quicksilver G4 single 800.

 

I'm trying to back up a machine running Panther 10.3 off a FW drive. Quicksilver dual-800. Client version 5.1.109.

 

Two previous backup sets have been successful over the past several months.

 

I started a new backup set on 11/7, that's where the trouble is happening.

 

It backs up the first partition of the client, doing 29.6 GB with 139814 files, and then gets to "Closing". Now it never ends, leaving 179 files taking 12.8 MB. I've tried repeating this by backing up clients in different order. Same problem on "closing": it never ends.

 

Today, I've waited 24 hours for it to finish "closing" and it still isn't.

 

Please help. Give me something to try.

 

I've done a recycle backup 3 times on this, and it always stops at this point. I'm desperate for a hint!

 

Using Grappler SCSI 930 to external Ecrix VXA-1 drive on the server. AFAIK, it has the latest firmware.

 

SC

 

P.S.

It first backed up 8 GB off a Windows client, then the first partition of the Panther client as I said above. Therefore, it went through one 33 GB VXA tape. I put in a 2nd tape when requested. Then it finishes backing up this partition, goes to "closing" and never gets out.

[scstr]

I've noted 2 things now. Apparently there's a Retrospect 5.1.175, so my version may not have the latest update. I also note a Power Mac SCSI update here:

 

http://forums.dantz.com/ubbthreads/showflat.php?Cat=&Board=Announcements&Number=31660

 

I'm reluctant to abandon a failed "closing" again for my backup, but I may have no choice. I'll wait a little while to see if any kind soul (Robin? ) answers my plea.

 

[scstr]

Update 2:

 

After 27+ hours, I finally Force Quit Retrospect.

 

I ran DiskWarrior on the server. 14 files had incorrect text-encoding values.

 

I booted into single user mode. fsck -y came up clean.

 

Then I downloaded ... installed 5.1.175 on the server, followed by the Power Mac SCSI update and the 4.1 driver update.

 

Retrospect launch now shows version 5.1.175 and Driver Update version 4.1.109 (where it used to be 4.0.103).

 

I then created a Server script with *just* the problematic G4 (and my 10.2.8 1 GHz Tibook after) to be backed up. Then I hit Recycle for the Backup set. AGAIN.

 

Then I started the Backup. My fingers are crossed. I've already spent a lot of time on this (half of my working hours this week), and I'm getting frustrated.

 

 

[MMF]

Hi,

Having the same problem. Dantz, where are You??? Not a good thing if you can't backup your software. How about a refund, maybe that will get Dantz's attention.

[AmyJ]

Quote:

---

Two previous backup sets have been successful over the past several months.

I started a new backup set on 11/7, that's where the trouble is happening.

 

---

 

Quote:

---

Then I hit Recycle for the Backup set. AGAIN.

 

---

 

Since the problem occured when you started a new backup set the place to start would be to create another new backup set, rather then recycling the existing one.

 

 

Quote:

---

Using Grappler SCSI 930 to external Ecrix VXA-1 drive on the server. AFAIK, it has the latest firmware.

 

---

 

Make sure you've tried different/new tapes. Clean the tape heads.

 

Try backing up the drive in question to a "File" backup set, rather then a tape backup set to further isolate the problem.

 

One last thing to note - Apple has confirmed a problem with Panther and some Firewire drives, which can cause data loss. Please see the following link for details to see if you may be affected by this:

http://www.apple.com/macosx/firewire800specialmessage.html

[arloleach]

I'm guessing this doesn't have anything to do with your tape heads or your backup set; I'm experiencing the same problem running a Duplicate operation to an external FireWire drive. Since upgrading to Panther, the Closing portion of the backup process runs very slowly. It will finish if I wait long enough! But, for comparison, backing up my Home folder used to take about 10 minutes and now takes about an hour; backing up my entire PowerBook used to take about an hour and now takes about ten hours. Before, during the Closing portion, the filenames used to flash by so quickly that I couldn't read them; now, each one appears for a second or more and I can easily read them. The worst part is that other applications slow down to the point of becoming barely usable during this process (the biggest delays occur when switching between programs).

 

I'm finding that about 3/4 of the time, I just cancel during the Closing process because I need to get some work done or pick up my laptop and go. I used to have no problems getting full, complete backups in a short amount of time.

 

So, here's confirmation of the problem; any ideas about a solution?

[arloleach]

P.S. I'm running Retrospect Express 5.0.238. I'd gladly upgrade to 5.1 if it addressed this problem, but since you're experiencing the same problem with 5.1.x, I guess that wouldn't help.

[charlesblackwell]

Yeah don't bother upgrading - closing time is appalling - apparently its do with permissions - why they can't make permissions backup optional, I have no idea.

[Mayoff]

With each duplicate process, Retrospect attempts to set the permissions for every file on the destination every time. If the destination contains 100,000 items....that can take a long time for the finder to complete. Hopefully we can speed this up in a future release.

 

Writing to a file backup set, should not have this type of problem.

[ilader]

I have encountered the 'endless closing' situation a few times. What has worked for me is to delete and rewrite the automated backup script, pointing the new script at the same sources and destinations as the old.

[arloleach]

Hmm, I wonder if ilader's problem is the same. I'll have to try that solution.

 

Also, if this is a permissions problem, as Mayoff explained, would using the "Ignore permissions on this volume" setting on the backup drive help? Would that still create bootable backups?

 

Better yet, now that version 6 has been announced, can anyone at Dantz tell us if that version addresses the problem? If so, I'll gladly buy upgrades for everyone at my company. (If not, I'll keep fielding tech report requests that say, "My backups never finish." )

 

-Arlo

[larryross999]

Version 6 did not correct the long closing problem for me, although mine is none near yours in length. It is taking me about 65 minutes to perform a duplication, with most of the delay in the 'closing' marathon. I thought that was too long, and it has been that way for several versions of Retrospect now for me, starting with version 5.0.

[arloleach]

Thanks for the report, lr999. I ran my first backup with a trial installation of Retrospect 6 last night and it seemed that the "closing" phase was not working any better than before; this confirms it. Backing up my user folder to an external FireWire drive used to take 10 minutes with no perceptible impact on system performance; since upgrading my OS to Panther, it takes over an hour and slows down all my other applications to the point where I usually just cancel the backup. I was ready to order Retro 6 upgrades for my company, but I'm glad I was able to check this in a trial version first; now I won't waste the money.

 

I've always used Duplicate so that I can easily pluck individual files off my backup drive if I need them, rather than having to go through Retrospect to carry out a restore operation. But I guess I'll try Mayoff's suggestion of backing up to a backup set. I'm also looking into alternatives to Retrospect, although to be honest I don't know of any that are in the same league.

 

-Arlo