DLT IV TAPES - PERFORMANCE

[David Ruska · Accepted solution]

Thanks for the logs. The DLT drive has a considerable number of cleaning and error rate related soft error messages. Here's some examples:

---
Event # 45 - Packet # 173 - SCSI Event: Cleaning required
Temp 37 C
POH/PC/MID/SK/ASC/ASCQ/AddErr=21242/35/78D71592/3/80/01/00
Event # 46 - Packet # 174 - SCSI Event: Cleaning requested
Temp 39 C
POH/PC/MID/SK/ASC/ASCQ/AddErr=21257/35/78D78703/1/80/02/00
Event # 47 - Packet # 175 - SCSI Event: Soft Error Exceeds Threshold
Temp 37 C
POH/PC/MID/SK/ASC/ASCQ/AddErr=21268/35/78D78703/1/80/03/00
---

These are within the last 100 hours (POH=Power-on-hours) of use.

The 20041001_tempjob_poortape.ltt file shows the following:

Drive Error Statistics
Read errors corrected w/substantial delays : 0
Total read rewrites or rereads : 0
Total read errors corrected : 0
Total read bytes processed : 0x0000000000119EA0
Total read uncorrected errors : 0
Write errors corrected w/substantial delays : 0
Total write rewrites or rereads : 4171
Total write errors corrected : 4171
Total write bytes processed : 0x00000000460EEA70
Total write uncorrected errors : 0

If you take the number of rewrites and divide by the total bytes (displayed in hex), you get a representitive error rate calculation. In this case it would be 4171/1175382640, which comes out to 3.5 rewrites per MB. That's poor error rate. Anything over 1 per MB will have a noticable hit on performance.

The other ticket, 20041001_Overnight.ltt, does not have error rate data. The drive status is "load command needed" which most likely means the host sent a command to unthread the tape (although the cartridge may still be in the drive).

The cause of poor error rate can be due to any of the following:
* Dirty head (normal staining & debris)
* Contaminated head (from environment)
* Worn/profiled head
* Electronics problems
* Worn media

If the head is stained from normal use, running a new cleaning tape through the drive 3 times in a row will typically clean it up.

Head contamination is harder to remove with a cleaning tape, although sometimes it helps. Some sources of contamination we've seen include in the past include smoke, soot/ash, paper dust (from printers), ink (from inkjet printers), carpet fibres, and outside dust.

Over time the tape drive head wears, since it is in direct contact with the media. It's therefore possible for the head to profile as a result, and that may make it read/write on one manufacturers formulation better than another. The tape stiffness and abrasitivity can vary within the allowable range from vendor to vendor.

The tape serial numbers you included in your email indicates that you have tape from two different manufacturers. There's a letter in the middle of the SN that indicates which one. You can see if your test results show one working better than the other.

Since you mention having problems with new tapes, we can rule out worn tapes as the cause. Note that new media may occasionally have higher error rate for the first few passes down the tape, as there is some debris that gets left by the splitting process, which typically cleans itself off with a fairly short amount of use.

If you conclude the head is worn, then the only real solution might be to replace it.

[Craig Clark_1]

Thank for looking at the log files. Thats a lot of information you've gathered.

I will try cleaning the tape drive as you suggested. Then I will run LTT tape test on the two different sets of HP media and save the log files.

If you would be willing to look at these two files and give me a final analysis, I would appreciate it.

Thanks you for your time so far in helping me.

[David Ruska]

Craig,

No problem. Just send them to ltt_team again, and post a quick reply here so I know to go look for them.

[Craig Clark_1]

Hi Dave

I have posted up the LTT support tickets. If your able to give me some advice based on your experience on the units that would be great.

With regards to the back up jobs the data rate has improved to 295MB/min average (not too far from the 315MB/min target).

At the moment I'm not sure if I should be looking to obtain a new unit or get the head examined/replaced.

I'm tempted to run the drive over the next two months with a two week cycle of tapes (one week new tapes, second week old know good tapes) and monitor the error rates as reported by LTT after each job. At the end of each week using new tapes use a cleaning tape (that would equal to 4 cleaning cycles over the two months, hopefully this is not excessive as I do not want to damage the heads due to over cleaning).

Hopefully this action will nurse the unit back to full health.

All that being said none of this is worth doing if the integrity is in doubt as the data is the most important factor. I have a SDLT offsite to do test restores with (I assume there is no risk of contaminating this unit?). All the tape drives are in normal office/home environments, whilst not clean room air filtered standards they are not smoke ridden dusty places.