Ultrium 1740 performance

[NoiK00]

Hi all, 

I recenty bought a new external ultrium 1760 since the old one we had broke down, leaving a tape inside. 

Since i attached the new tapedrive, i haven't been able to create a backup that completed in time, or most importantly can rely on. 

When i wanted to install the device, i received a cable that didn't fit with our setup. I receveid a SAS cable SFF-8088 to SFF-8644. where the SFF-8088 is on the side of the drive, and SFF-8644 would be on the server. I then bought a SFF-8088 to SFF-8088 cable, since i had multiple SFF-8088 slots free. Ever since, i didn't managed to get a decent performance out of it. 

The server is a Proliant 385 G6(i know, not the newest) where i am running backup exec. but when testing the performance with HP L&TT there already is a warning: 
Drive Performance test passed. Drive performance is constrained. Check HBA and cable configuration. Also try another tape and clean the drive.
See www.hpe.com/support/pat for more information.

I already updated all the drivers to 2016.10.0 without result; 

Does anyone has some more advice for me trying to restore a backup capability with decent performance?

Thanks in advance, Kees

[Curtis_Ballard]

What 8088-8088 cable did you buy?

SAS tape drives are very sensitive to good quality cables.  I did some work on SAS cables back when 6Gb was under development and we found a lot of cables that had poor solder joints inside the connectors which caused retries on the SAS link and typically when that happened it caused a lot of retries and performance was poor.  The suppliers we approved for HPE SAS cables were the best we could find and we rarely have any issues with those cables though I have seen a couple over about 8 years.

If you post an HPE LTT support ticket we can determine whether there are SAS retries occuring.

[NoiK00]

Hi Curtis_Ballard, 

Thank you for your reply. 

I bought a "STARTECH SFF-8088 to SFF-8088 2M EXT SER SAS". Startech Vendor product number: ISAS88882. I have no clue what the quality cable is. Are these cables also sensitive when using around corners?

I also created a *.LZT, do you mean this with a " HPE LTT support ticket"?

It is downloadable from this link: https://1drv.ms/u/s!AuCiSmd0WMX3mS5o69nnNty8F-lr

Thanks in advance, 

Kees

[Curtis_Ballard]

I can't say for certain who makes the Startech cable but I took a quick look at the specs and it says it is TAA compliant which narrows it down a lot.  None of the cable vendors where I've seen a lot of problems were capable of making TAA compliant cables.  TAA compliance isn't directly related to quality but I know the cables that I had trouble with weren't capable of being TAA compliant.

The logs you pulled are the correct ones.  Nothing in the logs indicates a cable problem so that probably isn't it in this case.

Internal write performance all looks really good.  Reads performed a few low level corrections but not enough that there should have been any visible performance degradation. The level of corrections wasn't too bad, there is always some ECC activity.

The performance measurements do show performance degradation with the drive running less than 1/2 speed.  The logs show that the drive keeps pausing waiting for SCSI activity.  That usually means either bus problems causing low level retries and holding off activity or something in the host causing slow SCSI command send/receive.  Since the ticket doesn't show any bus level problems there might be something at the host end. 

A 385 G6 should be able to run much faster than that.  You already used the LTT performance test and it shows performance problems that aren't caused by Backup Exec so it is something below that.  Have you checked that your HBA firmware and drivers are all up to date?

[NoiK00]

Hi Curtis_Ballard, 

Thanks again.....

So, you're saying the cable should be alright, check the firmware and drivers. Then the trouble starts for me....Like said, i recently used the SPP 2016.10.0 to update the server. 

The trouble start when you ask me for the versions. am i looking in the correct places??

The driver version i see in device management: 

"Storage controllers">"LSIAdapter, SAS 3000 series, 8-port with 1068E" - version 1.28.3.52 (date:7-7-2008). 

When opening Insight diagnostics, i see Firmware revision 1.28.2.0 on both the "LSI adapter, SAS 3000 series, 8-port with 1068E". 

Thank you,

Kees

[Curtis_Ballard]

I can't tell if those driver versions are the recommended versions for tape.  The 1068E is too old.  I only have the last 5 years of support database information and we quit supporting that controller before that.  You might check Broadcom and see if they have any drivers for that card but it may be too far back in history.

I don't recall the 1068E being a Proliant option so the Proliant driver pack may not have updated it.  That controller was supported with HP tape as it was used in HP Workstations.

The drivers may not be the problem but it is always good to check.

It can be hard to figure out performance problems if nothing seems to be generating errors.  A few things I can think of that you might check:

o load a tape and do some read/write testing then preferrably before unloading or shortly after unloading pull the drive logs again. The last set of logs was pulled after a tape was loaded but no reads/writes to that tape had happened so there were some empty logs that might have some useful information.

o look in the Windows event log for any LSI events logged during your read/write testing.  If you find any post the full information here. I have the engineering documentation on the LSI drivers for that controller and LSI put some really useful information in their events.

o check the Device Manager and see if there is an option to roll-back the driver for the SAS controller.  Once in a while a newer driver will introduce a problem that wasn't there before.

o can you think of anything that changed since your last drive?  You mentioned needing a new SAS cable.  Did you change the controller, attach to a different server, change to a new OS?

[NoiK00]

Hi Curtis Ballard, 

I didn't had a rollback option in the device manager, so I just reinstalled another driver: LSI Adapter, SAS 3000 series, 8-port with 1068E -StorPort - 
1.26.5.0. It is a downgrade and a StorPort version. After rebooting the server i logged a ticket, performed a installcheck and logged another ticket. I hope it contains the data you were looking for?

Link: https://1drv.ms/u/s!AuCiSmd0WMX3mTHj3awokret0fnl 

Additional i added a part of the systemlog since the boot, till completing the installcheck. it contains some LSI errors/events like you expected. 

Since our previous tapedrive failed, we purchased this new one and later the cable no other changes are known(except some windows update ones..). Only after experiencing these problems we started with the SPP updates. Like said before, i am only using a different controller. The previous one where the old tapedrive was connected, was a LSI Adapter, Ultra320 SCSI 2000 series, w/1020/1030. Now we are using the 3000 series. 

Thanks, 

Kees

[tapelibraryfixr]

How fast are actual backups?

[Curtis_Ballard]

Sorry but I get an error for link not found trying to open the link for the latest logs so I'm not able to see if there is any new information.

I might not get a chance to check back here for a few days but will keep watching this spot.

[NoiK00]

Hi Curtis Ballard, 

Sorry, here i got a new link: https://1drv.ms/u/s!AuCiSmd0WMX3mTEn0w7oB_9TZS8d

I yesterday started the jobs again, started a job for a server with about 500 gig of data. the Jobs was running from 18.00 yesterday till now. The speed is 5 MB/Min. 

[NoiK00]

hi all, 

i updated the driver through Windows, it installed a "LSI Adapter, SAS 3000 series, 8-port with 1068E" again, with version 1.28.3.52. The L&TT install check now states another error: 

"

Drive Read/Write Test :

Read/Write test version: V15.04.2015
Drive Read/Write test failed.
Dataset not in file: LTT:WRITBACK
Write Back data written to device at path
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Drive Performance Test :

Drive Performance test failed. It could be that the media is faulty. Try another cartridge. If the problem persists the following resources are available:
Drive Performance test aborted, drive failed to load the media. HPE Library and Tape Tools from the CD or from www.hpe.com/support/tapetools. Run the device analysis test.

"

Were using the same test tapes we used before, that were new tapes fresh out of the plastic. 

After exploring the HPE L&TT program i saw we were using an older firmware version. The Check says we could update tot Revision U64D. I am not sure if this will fix it...But why does a a new drive(2017) still needs a firmware from 2011?! Was it on the shelf all the time?! Too bad i haven't got a support contract with HPE so i couldn't update it...Any ideas if this will do anything for us?!

Thank you, Kees

[tapelibraryfixr]

its looks like either a drive or sas card problem. just have to go through a process of elimination.

[Curtis_Ballard]

I think that the errors you got from LTT are mostly an example of poor messaging.  I've provided feedback to that theam several times that their messaging is poor.  In this case there is the statement "drive failed to load the media" which suggests that there was a problem before the real test even started.  The test does give you good advice though, manually load a tape then use LTT and run the "Device Assessment Test" - that will do a lot of read/write and other activities along with low level log analysis and is a pretty good test for drive/media problems.  Sometimes it has to be run twice if it detects problems that could be either the drive or the media.

I looked at the logs you posted but they weren't conclusive.  The Install Check just doesn't do enough read/write activity to get good log data but there were some things in the logs that looked like the drive may be having trouble writing. The rate of putting data onto the tape after the data was in the internal buffer was low but there wasn't enough data written to be able to cover for the normal mechanical start/stop time that has to occur at the start of writing.

I reviewed the Windows system log files briefly, I didn't spend a lot of time digging but ran a quick filter and the viewer reported that there weren't any LSISAS entried in the event log so if the controller is having trouble it isn't logging anything.

I think the Device Assessment Test inside L&TT is your best option.

You asked about the firmware, the drive logs show that the drive is running U64D so I don't know where an indication that the firmware should be updated might have come from.  There was one build after U64D but I reviewed the change list and nothing would help you.  The primary reason for the later build was specific to tape libraries so that drive firmware likely was never released for standalone drives.