Jump to content

RMS freezes *hard* during network backup


awnews

Recommended Posts

I'm having a serious and annoying problem with an installation of Retrospect MultiServer 6.5.350 [Value Pack, so OFB, Proactive, etc.], RDU 5.3.103. Running on a Windows 2000 Server, SP4 PC (450MHz Dual PIII, 512MB PC100 SDRAM). I'm backing up with a Client on another W2K Server PC, backing up the "C" or "D" partition of this SCSI drive (NTFS) to an 80GB Ximeta Ethernet drive (NTFS) on a 100M LAN running the 3.08 driver. I'm running under the "Administrator" account (with admin password) on the machine as is Retrospect (always left running, no Launcher service).

 

If I back up the "C" partition on the other drive to the Ximeta it works fine.

 

If I back up the "D" partition, Retrospect craps out--hard--in the middle of the backup and runs no other backups. Part way thru the "Copying" phase (variable time and file point) Retro just hangs and stop doing anything (for hours or days). Note that this is not a "stuck in closing" problem since it never gets that far. The GUI becomes completely non-responsive and I have to resort to the Task Manager to kill the program. However, this *only* kills the GUI as "Retrospect.exe" remains running on the machine. If I try to kill the still running "Retrospect.exe" via the TM, I get the message "The operation could not be completed. Access is denied." So I then have to reboot the machine to restore normal operation.

 

The backup file has become corrupted at this point so I have to do a Re-Catalog to repair it. This works (no errors reported, a single .rbf at the end) but the process repeats the next time this backup runs *or* if I attempt to run a new backup of this type from scratch.

 

Looking at the Retrospect log reveals nothing. The log doesn't even show that this job started or ran. After the reboot, the Event pane shows "Execution terminated unexpectedly, possibly due to a power failure or system crash" caused no doubt by the fact that I had to force a reboot of the machine.

 

I don't "mind" if Retrospect is unable to back up this fileset to this drive, but it should *not* hang and prevent other scheduled and proactive backups from running.

 

-------------------------

As another datapoint, I'm running Retro Pro 6.5.350, RDU 5.3.103 on two other XP Pro machines, also with a 120 Ximeta Ethernet drive on a 100M LAN and the Ximeta 3.08 driver and am not having any problems backing up to the drive.

Link to comment
Share on other sites

Hi

 

 

 

How big is the backup file when this failure occurs? Does the error occur if you recycle the set and backup the D drive first? I wonder if the same problem would occur with a disk backup set?

 

 

 

Are there any really large files on the D drive? If you break the D drive into subvolumes and back them up individually can you narrow the problem down to a specific file or area of the disk?

 

 

 

If its just a problem accessing a file Retrospect should just log an error and carry on. However it looks like this may be more serious...

 

 

 

Thanks

 

Nate

Link to comment
Share on other sites

On three tries (scheduled backup, with freeze and corruption of main File backup) I saw the progress bar frozen all over the map (% of copying, file at point of freeze, etc.). When I tried it yesterday (try four), I tried starting a new backup set and the (Copying) progress bar was frozen with only 111KB copied out of just under 10GB that needs to be copied. So it's not the size of the destination. And based on this I have a hunch that a Disk backup wouldn't do any better.

 

I'm also backing up this same D partition to another harddrive on the same system and that hasn't frozen yet. It's only when I try to do the backup over the network to the Ximeta. But I *am* able to backup the C drive to the Ximeta without a freeze (so far).

 

*One* of the freezes happened during a copy of a 51MB file (modestly large). I'll try zipping that and see if it makes a differences. I'll also trying running Norton Utilities (not installed on this server at the moment) to see if it can find any disk or file problems.

 

At the end of the day, I'm more concerned about the *hard* way Retrospect freezes--no more progress (doesn't even give up on the in-progress backup that has failed), some sort of "disconnect" between the GUI and the running app such that I can't TM kill the program, etc. Failing on a file, unable to copy more data, etc. is "fine," but the app freezing is not...

 

And, as noted, there's *nothing* in the log during the frozen run (not even an indication that the backup was started). I would think it would make sense to log messages as things are running (starting backup, got to there, did this, etc.) rather than wait until completion so that there would be some record of what was going on in case something went wrong.

Link to comment
Share on other sites

Agreed - the type of hang is troubling. Although I have a feeling this is something beyond Retrospect's control. As a result it isn't handling it very well.

 

I don't think file size is the issue. You should have zero problems with files under 1GB. I also agree that a disk backup set will probably have the same problem - worth a try though...

 

So some network backups to the ximeta disk work but others do not. And Retrospect freezes solid to boot. Interesting.

 

I would try a couple things:

-Try a backup with open file backup turned off

-Slow the speed of the network driver to 10MB and run a test backup.

 

Are there any errors in the Windows log when this hang occurs?

 

Thanks

Nate

Link to comment
Share on other sites

  • 2 weeks later...

New info:

 

1) It turned out I had driver 3.07 rather than 3.08 installed for the Ximeta. I don't know if it made any/the difference and Ximeta doesn't really list differences/changes/improvements for driver versions (about the same level of detail as an RDU update...), but it wasn't *the* latest version.

 

2) I tried a "Disk" rather than "File" backup and get the same "Freeze-up" failure (on a different file).

 

3) I had disabled backup of the "D:" partition (via Client backup on another PC) for the time being, but last night I experienced the "freeze-up" failure for the first time when running a scheduled backup of the C: drive of the PC running Retrospect. So the problem is *not* just associated with the particular D: partition.

 

4) I had considered reverting to an earlier Ximeta driver (3.03) but Ximeta *just* released a driver update, 3.10. So I decided to try that first (with File backups of C: & D: via a client, C: locally ==> Ximeta drive) and see how that goes. I'll report back with my findings in a few days.

 

I have *not* tried either of Nate's suggestions (turn off OFB, slow network to 10M) since these backups *used* to work without these options (maybe when using the 3.03 driver?). And no, there were no messages in any of the Windows Event logs *or* in the Retrospect log (no log of hung backup even being run let alone any problems reported).

Link to comment
Share on other sites

Hi

 

Just to clarify a couple of points:

The error only happens when backing up data from a client machine over to the Xmeta network disk? In other words:

 

Backup client->backup set on local disk of the backup server=OK

Backup local disk data->Backup set on Xmeta = OK

Backup client->Backup set on Xmeta = Fail

 

If that is the case I suppose a network driver on the backup server could also be a point of failure. Are there any updates available for that NIC?

 

About the log info: By default Retrospect keeps log info in a temp file until an execution completes. This prevents keeps the final log in order even when you run multiple executions. In the hidden preferences you can tell Retrospect to flush log on write. That forces Retrospect to update log info in real time.

 

Thanks

Nate

Link to comment
Share on other sites

> Backup client->backup set on local disk of the backup server=OK

*** True, backing up from clients to local IDEs (not Ximeta) never freezes.

 

> Backup local disk data->Backup set on Xmeta = OK

***** I've now seen one FAIL here *****

 

> Backup client->Backup set on Xmeta = Fail

*********** TRUE *********

 

> If that is the case I suppose a network driver on the backup server could also be a point of failure. Are there any updates available for that NIC?

 

*** Don't know, but nothing shows up with Windows update. But I don't think this a NIC issue. I'm backing up 50+ clients over a 100M/10M network to local IDE harddrives (not Ximeta) without errors. It's only when backing from from a source (a client on another PC, the local PC) *to* a network-based Ximeta 3.07 where I've seen Retrospect experience the hard freeze.

 

 

> About the log info: By default Retrospect keeps log info in a temp file until an execution completes. This prevents keeps the final log in order even when you run multiple executions. In the hidden preferences you can tell Retrospect to flush log on write. That forces Retrospect to update log info in real time.

 

*** Depending on what I see with the new Ximeta 3.10 driver, I may turn that option on so I might have a chance to see *something* in the log on the failure.

Link to comment
Share on other sites

So last night the backup:

 

S) Retro 6.5.136 Client running on SERVER1 (W2K Server, SP4), D: NTFS partition is backup source

 

D) RMS 6.5.350 [assorted RDUs used, latest 5.4.110] running on SERVER2 (W2K Server, SP4), Ximeta driver 3.10 installed,

==> 100M Ethernet-connected Ximeta 80G NTFS drive is backup destination

 

was able to run error-free w/o freeze to completion (File backup), for the first time in a couple months.

 

 

The history *seems* to have been

 

1) Ximeta driver 3.03 ran OK

2) Ximeta driver 3.07 caused/interacted-with Retro to freeze up (non-responsive, GUI could be TM quit but ....."Retrospect.ext" continued to run and couldn't be TM killed, had to reboot to recover then recat File)

3) Ximeta driver 3.08 never tried and unknown reason for +0.01 update

4) Ximeta driver 3.10 ran OK

Link to comment
Share on other sites

  • 2 weeks later...

I wasn't aware that the Ximeta drive is using SAMBA. Do you have info on the internals? From the PC's perspective, there's a custom Ximeta driver that gets installed on the PC since the Ximeta uses a non-TCP protocol ("LPX") over the LAN.

 

LPX - Lean Packet Exchange protocol - This is the new ethernet protocol designed by XIMETA that is used to communicate with the NetDisk. It provides both security and speed for NetDisk users. [note: it won't route]

Link to comment
Share on other sites

Archived

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

×
×
  • Create New...