Jump to content

Backup to tape frozen


Recommended Posts

I'm doing the usual Normal backup to tape(every night at 12AM). When I walked in today at 9AM I see:

 

Copying projects on DataHD

Remaining: 72.7G

Completed: 210.6G

>CHA_4486.JPG

 

On the devices window I see:

3-Weekly catalog[056] writing

 

4: 2-Weekly Catalog [057]

5: Empty

6: 1-Weekly Catalog[056]

7: 3-Weekly Catalog[056]

8: 2-Weekly Catalog[056]

 

Should I erease all tapes and have Retrospect try again tonight? We do a New Media backup every Weekend and the current one is a normal backup. Any help is appreciated!!

Link to comment
Share on other sites

Is the server operational otherwise? What does "ps -axl" show for the retrospect process ? Can you ssh in to the server?

 

Note that your RDU (retrospect Driver Update) is a bit out of date. Current version is 6.1.9.102, and that might help this issue. Updates are here:

Retrospect Mac updaes

 

Sounds to me like a communications problem with your tape drive or the interface to which it is connected. You haven't provided any information as to how your tape drive is connected to the Xserve (SCSI? Firewire?) or what card(s) are used, or whether there are any other devices on that channel.

 

I think that there may be some issues with the 3-Weekly catalog[056] member as a result of the hang (missing EOT, etc.), but the other members should be OK. I'd also wonder about the integrity of the catalog for Weekly Catalog [056] backup set.

 

But nothing you have mentioned indicates that there is a fundamental problem with the tapes, or that all tapes need to be erased. Yours seems to be a communication problem with the tape drive.

 

We had a similar issue a couple of years ago, where Retrospect would hang when doing its initial scanning of the devices if there was a disk drive on the same SCSI channel. I believe that Retrospect was causing some sort of a bus reset that caused a disk transaction interrupt to be dropped. Workaround was to put the tape and autoloader on a SCSI channel with no other devices. Our setup is an Xserve G5 with ATTO UL4D and Exabyte VXA-2 1x10 1u PacketLoader (SCSI), happened on multiple versions of Retrospect 6.0.x and 6.1.x, multiple versions of Mac OS X Server (10.4.x).

 

Although I don't think it is related because you say you are backing up to tape, not disk, it's unclear what the disk drive is that is in your LaCie d2 case (there are multiple drives that are used there). Apple System Profiler would tell you.

 

Russ

Link to comment
Share on other sites

I have the exabyte tape machine connected using a firewire, hard drive(dataHD) and external LaCie drive. At this moment we are only using the tapes to backup data from the dataHD which is the share drive for our people to store important projects. If I update the RDU do I need to restart the server? Right now I'm erasing all tapes to see if tonight there are no errors. Should I erase catalog [057] and use [056] or should I go with [057]? There was only one tape with [057]. Should I erase [057]? Thanks!

Link to comment
Share on other sites

I'm a little puzzled by the tape listing. It shows 3 members of backup set "Weekly Catalog[056]" and one member, the second member, of backup set "Weekly Catalog[057]." How did that tape member get its name? Did Retrospect assign it, or did you give it the name manually? What happened to member 1-Weekly Catalog[057]?

 

The purpose of a new media backup is to enable you to retain the previous backup set(s), while using you same scripts to back up to the latest backup set. When the new media backup is performed (or if you manually switch to new media in Configure> Backup Sets), Retrospect should automatically change all scheduled backups in all backup scripts to use the latest backup set. Thus it seems especially puzzling that your script should be backing up to Weekly Catalog[056] instead of Weekly Catalog[057].

 

If you go to Automate> Preview, what backup set is called for in the next scheduled backup events; is it Weekly Catalog[056] or Weekly Catalog[057] or some combination of the two (which shouldn't really happen)? ]f you go to Configure> Backup Sets, is there a listing for both Weekly Catalog[056] and Weekly Catalog[057]?

 

You probably did not need to erase backup set Weekly Catalog[056]. If the hang happens in the future, try going to Tools> Repair> Update existing.

 

Please report back to us with the above information, and in the meantime, I wouldn't mess further with any of your backup sets.

Link to comment
Share on other sites

This is an error I found in the logs:

 

New Media Backup using Weekly Backup at 4/15/2007 1:00AM

To backup set Weekly Catalog [056]

4/15/2007 1:00: 17AM: New Media backup: The backup set replaced with Weekly Catalog [057]

 

4/15/2007 1:00: 17AM Copying projects on DataHD

>Can't read file "somedir/somefile", error -43(file/folder not found)

 

So what I did yesterday I erased all tapes and modify the backup to use Weekly Catalog [057] instead of [056]. The backup is supposed to happen everyday at 12:00 (not the new media, which is on Sundays). Today when I walked in this morning I see that the backup still running and it is still running at this moment. Why is taking so long? It started at midnight and still has 32GB to go. thanks

Link to comment
Share on other sites

I'm still puzzled. The log section you quote indicates that the new media backup was run on Sunday 4/15, which shifted the backup set to Weekly Catalog [057]; yet you said earlier that the script that was running on Tuesday evening was backing up to Weekly Catalog [056]. Was that really the case?

 

Do you use the same script for your weekend new media backup as for your weekday normal backups?

 

It's unlikely that the error you note in the Sunday log has anything to do with the other issues you've experienced.

 

Quote:

Today when I walked in this morning I see that the backup still running and it is still running at this moment. Why is taking so long? It started at midnight and still has 32GB to go.

 


What does the backup window (or if the backup has finally completed, the log) say about the performance rate in MB/min? Is it in the copy phase or the compare phase? How many GB have already completed?

Link to comment
Share on other sites

Quote:

I have the exabyte tape machine connected using a firewire, hard drive(dataHD) and external LaCie drive. At this moment we are only using the tapes to backup data from the dataHD which is the share drive for our people to store important projects.

 


Unclear from this whether either or both of your hard drive ("dataHD") and/or external LaCie drive are firwwire. If either is, and if they are on the same firewire chain as the Exabyte, then this may be the cause of your original "hang". You still haven't answered my original questions as to whether this was really a "hang":

 

Quote:

Is the server operational otherwise? What does "ps -axl" show for the retrospect process ? Can you ssh in to the server?

 


You really need to investigate the cause of that "hang" because it's not normal and might happen again.

 

Quote:

If I update the RDU do I need to restart the server?

 


No, but you need to quit Retrospect, copy the new RDU into Retrospect's folder, then restart Retrospect.

 

Quote:

This is an error I found in the logs:

 

New Media Backup using Weekly Backup at 4/15/2007 1:00AM

To backup set Weekly Catalog [056]

4/15/2007 1:00: 17AM: New Media backup: The backup set replaced with Weekly Catalog [057]

 


This is not an error. Your schedule indicated that a "New Media" backup was to be done, moving from backup set "Weekly Catalog [056]" to the next backup set named "Weekly Catalog [057]"

 

Quote:

4/15/2007 1:00: 17AM Copying projects on DataHD

>Can't read file "somedir/somefile", error -43(file/folder not found)

 


This is more informative than perhaps an error, and might not indicate any problems. Since you have obfuscated the error message by editing the directory/file name, it's not possible to be specific as to what happened, but I can speak in generalities. All it means is that, between scanning and pushing the files out to tape, some file disappeared. If your system was active (rather than quiescent), this is normal. You seem to indicate from the information that you have provided that this is being done on an Xserve running Mac OS Server 10.4.6. You might, for example, have the mail services active, in which case email messages will be coming and going, and, as they move through the postfix/amavisd-new/ClamAV/SpamAssassin chain, files will be created and deleted in the process. Likewise, since it is a server, users will be creating and deleting files as you are doing the backup. Etc., etc. Unless you shut down all services while you do the backup, such errors (informational notices) will always occur. It's normal.

 

Quote:

So what I did yesterday I erased all tapes and modify the backup to use Weekly Catalog [057] instead of [056].

 


It was not necessary to erase the tapes. By doing so, you lost all of your backup history on those tapes. Hope it wasn't needed.

 

Russ

Link to comment
Share on other sites

The backup was completed successfully. Tonight I need to have new media for the New media weekly backup which I believe will be [058]. Twickland the server is operation and everything seems ok with the server. The DataHD is an internal HD, the LaCie is an external (firewire) HD. Exabyte is on a separate chain. For now this problem seem to be solved. What I have been trying to accomplish is the "emailing notification" using the python script, because I was told here in the forums that it was more effective than the others, but no luck yet!

 

thanks for your help..

Link to comment
Share on other sites

Quote:

What I have been trying to accomplish is the "emailing notification" using the python script, because I was told here in the forums that it was more effective than the others, but no luck yet!

 


It's not that the Python script is more "effective", it's just easier to set up and keep working because Apple changes the interface to Mail with various releases, but the interface to Python (not an Apple product, but a defined language/system) stays the same across Apple releases. You do realize that the "Retrospect Event Handler" AppleScript for use with the "macmail.py" companion script is different from the "Retrospect Event Handler" AppleScript for use with Apple Mail ? Just want to make sure you are using the right set of files.

 

I've replied to your email questions with a cookbook approach in this thread:

Emailing Backup Reports

but haven't seen a recent response from you. My suggestion as to the easiest way to get the email notification working during testing/debugging is to define some folder on the server as a "subvolume", put a file in it, set that as the source, set the destination as a file backup, schedule a script about 10 minutes in the future to do that backup to the file backup set. That way, it will run very fast and provide quick turnaround during testing, and then you can simply delete the test backup from the schedule after you get email notification working.

 

If you still have problems, drop me an email (so I have your email address without it being posted in the forums) and I will email reply to you our known-working script with a discussion of our changes and installation instructions. But it really should be straightforward from my explanation in the other thread.

 

And again, your RDU is somewhat out of date but it wouldn't be the cause of the email issues. There was another RDU update this week (RDU 6.1.10.100), and it is here:

Retrospect Mac updates

 

Russ

Link to comment
Share on other sites

Archived

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

×
×
  • Create New...