The Error 203 (Hardware Error) Chronicles

[c3768c36-bf22-4c6d-bc39-cdddd3062225]

Here's the set up: G4 867, OS X 10.4.11, VXA 320 via SCSI, Retrospect 6.1

This exact set up was in use for a couple of years prior to the issues described below.

 

We attempted to upgrade to Retrospect 8 on a G5 with a brand new ATTO SCSI card. Of course, we weren't able to see clients because we were running V.8 on a G5 with 10.5.8 - at least that seems to be the consensus from the forums. So, we're patiently waiting for an update to address this issue. In the meantime I tackled two possible solutions. Moving V.8 with the ATTO card to a Mac Pro, or breaking out the old G4 setup.

 

The Mac Pro was a bust - the V.8 Engine would quit unexpectedly during any backup. So back to the G5 we go while I contemplate using the G4. At this point the common denominator between the Mac Pro and the G5 is the SCSI card - which had initially given me problems even showing up in the PCI bus, until I switched to a different cable (the previous cable and VXA show no problems when on the G4). So I'm crossing my fingers that this a problem with the card.

 

The G4 had been wiped to be put on Craigslist. I reinstalled Retrospect 6.1 on a clean system, brought over the catalogs that we had started at the first of the year - which had not been successfully updated in months due to troubleshooting these issues. Rebuilt my scripts, etc. My main priority is to clean off our Completed RAID - this drive stores all of our completed projects, when it starts to fill up we dump the older or less accessed files to tape - twice. One version stays here for our use, one goes offsite. The RAID is connected via Firewire to the aforementioned G5. A completed folder on the RAID is shared allowing all the artists to dump completed projects to the RAID. Our drives are getting full - we need to move stuff off that RAID to tape so we can clear off completed work from our drives.

 

The set up is simple - backup the entire Completed shared folder. We're up to about 350GB that needs to be on tape at this point. Here is where the Error 203 began to occur. About 50 or 60GB in it would stall on a large file - my r/w time would drop from 800mb/min to less than a hundred. And then I'd get the error. Of course this makes the catalog out of sync. Attempting to repair the catalog never finishes. It rewinds the tape, reads about 5 files, sets on "resyncing (slow)" … for hours. I've even let it set an entire weekend. And it will not sync. Fortunately we're only 2 tapes into this particular set. And I have a backup of the previous version of the catalog file - so I revert back, like nothing ever happened. Try it again. Different tape 3 (clean, straight out of the cellophane). Error 203.

 

Frustrated, I move the tape drive back to the G5 with 10.5 installed. This machine has both Retrospect V.6.1 and V.8 - not running concurrently of course. So I go through the same motions using V.6 - same error. And it happens on the same file as before. So I check the files - nothing seems to be up - they're not corrupt, I can copy them to another machine, open them up. So I think maybe this tape went bad during the previous attempt on the G4 - if it's writing to the same piece of tape that it was on during the previous error maybe it messed up the tape. So I try again - keep in mind this means reverting back to a previous version of the catalog and waiting for it to get through 60 GB data, I have real work to do and we need access to the tape drive to pull jobs, so I'm only able to do this troubleshooting sporadically.

 

The second attempt on the G5 fails with Error 203. In desperation I fire up the Retrospect Engine and try V.8 - keep in mind, this is a local share, we're not even accessing a client here, so I had hoped that V.8 would at least let me accomplish this. It gets to about 120GB - then Error 203. This was also on a clean tape - at this point I'm convinced that I either have a bad batch of tapes or that Error 203 is corrupting the tape I'm trying to write to. I'm sitting on 3-4 possible bad tapes.

 

So, I'm getting the same error on two different macs, the common denominators are (1. tape drive (2. scsi cable (3. files.

 

I run the Exabyte tools on the VXA - Full Tape r/w test. It doesn't even hiccup. Double-check the tape drives firmware is up to date. Order more tapes. Rinse and repeat on the G4 as well - no problems from Exabyte tools.

 

At this point I'm at a loss - I've moved the VXA back to the G4 - this is the set up that had worked for years prior. Except the cable is different, and of course the files. This time I get through the same 130+ GB as I did using V.8 back on the G5, it doesn't crap out around 60GB. I'm on about 30GB of tape 4 and Error 203. With the catalog out of sync, tape 3 is now crap. So decide to split the files up. If I can get half of the data backed up (which I now know I obviously can) then I won't lose as much ground when I get the inevitable error.

 

It worked - sort of. I got through tape 3 - over 160GB of data - I move this stuff out of the Completed Folder and move the remaining 160+ into the folder for back up. Error 203. Twice. On the same file, on two different tapes. I'm down 6 tapes now. The boss is not happy, this is getting expensive and very time consuming. If this is file related, why did it get past the previous file it was stumbling on before I tried V.8. Is it even possible that a file can cause the 203 error? Can that error corrupt my tapes?

 

I'm going to move the four instances of the "bad" file out of the backup and try again - but any other suggestions and/or recommendations are welcome.

 

Sheesh.

[twickland]

We attempted to upgrade to Retrospect 8 on a G5 with a brand new ATTO SCSI card. ... At this point the common denominator between the Mac Pro and the G5 is the SCSI card - which had initially given me problems even showing up in the PCI bus, until I switched to a different cable (the previous cable and VXA show no problems when on the G4). So I'm crossing my fingers that this a problem with the card.

What ATTO card are you using? Are you attempting to use the same card on the G4, G5, and Mac Pro? That may not work; the UL5D is only for a PCIe bus, while the UL4D is for PCI and PCI-X.

 

Given that you already experienced problems with the SCSI, I think this is where I'd start digging. Have you updated to the latest ATTO Config Tool, firmware, and driver? Tried different cables (easy to mash a pin on those LVD SCSI connectors)? Is your SCSI chain terminated properly?

 

Retrospect will push SCSI more than the VXATool; you may have a bad card.

 

I seem to recollect hearing about a bad batch of Exabyte tapes a few years back, so I suppose it's conceivable this could be your issue.

 

 

[c3768c36-bf22-4c6d-bc39-cdddd3062225]

What ATTO card are you using? Are you attempting to use the same card on the G4, G5, and Mac Pro? That may not work; the UL5D is only for a PCIe bus, while the UL4D is for PCI and PCI-X.

 

Given that you already experienced problems with the SCSI, I think this is where I'd start digging. Have you updated to the latest ATTO Config Tool, firmware, and driver? Tried different cables (easy to mash a pin on those LVD SCSI connectors)? Is your SCSI chain terminated properly?

 

Retrospect will push SCSI more than the VXATool; you may have a bad card.

 

I seem to recollect hearing about a bad batch of Exabyte tapes a few years back, so I suppose it's conceivable this could be your issue.

 

The UL5D is in the G5 (PCIe) - a version prior to the UL4D is in the G4 (PCI-X). Everything is terminated, the UL5D has been updated and I have been able to access it via the ATTO Config tool without any problems. I have yet to switch cables - only because I am running another backup currently on the G4 using the first batch of files - so far, no errors, this is on a tape that the previous batch had given me an error on the same file twice. Next I'll add the second batch back into the mix minus the files that appear to be giving me trouble.

 

I just can't believe that a couple of files would cripple the whole backup like this. I've been running Retrospect since the 4.0 days and have never had a single file or batch of tapes give me so much trouble. Thanks for the input though. If I hadn't experienced issues on two different SCSI cards I would've thought it was the card as well.