What's causing this?

[Fritz]

OS - Windows Server 2012 R2
MB - Supermicro X9SRH-F
HBA - Dell Perc H310 plugged into a 2.0 slot (same behavior in a 3.0 slot)
HD's - 6 IBM branded Seagate 2TB SAS 3.5



I've tried 2 different H310's and get the same results. I have other H310's in other boxes that don;t do this.

TIA

 

[i386]

I have other H310's in other boxes that don;t do this.

Are they also running server 2012 r2?

If not my guess would be windows caching mechanisms. Try to run the bpa on the server and see if it finds something unusual.

 

[Fritz]

Yes, the other re all running 2012 R2.

Are they also running server 2012 r2?

If not my guess would be windows caching mechanisms. Try to run the bpa on the server and see if it finds something unusual.

Ran bpa and got 2 warnings. Resolved both of them. One was disable 8.3 file naming and the other was set srv.sys to start on demand, I did. Still no joy.

 

[Fritz]

Wanna add I tried to load TrueNAS on this system and I could not get it to work. It would load with the HD's disconnected from the H310 but when I plugged them in all I got was a 20 minute screen of gibberish that ended in a load failed system halted message.

All drives test perfect according to HD Sentinal and Windows see no problem with them other than what listed above.

 

[Stephan]

With 6500 objects might be fragmentation bogging down transfer speed due to I/O on the disks. See if you have too many shadow copies on the volume. If this is over network, qualify network using netio or iperf first to exclude cabling and switches. Then copy a large single file, like a ZIP with a bunch of 5 GB Windows 10 ISOs thrown into it. Should be at least 10x size of RAM. Check reallocated sectors and other SMART attributes on all disks to see if they are still healthy or one disk is dying.

 

[Fritz]

With 6500 objects might be fragmentation bogging down transfer speed due to I/O on the disks. See if you have too many shadow copies on the volume. If this is over network, qualify network using netio or iperf first to exclude cabling and switches. Then copy a large single file, like a ZIP with a bunch of 5 GB Windows 10 ISOs thrown into it. Should be at least 10x size of RAM. Check reallocated sectors and other SMART attributes on all disks to see if they are still healthy or one disk is dying.

Thanks. Source is the boot drive which is a DC 3500 480Gb. I'm copying the contents of the download folder as a test. I'll try a large file and report back.

 

[Terry Wallace]

Another odd thing to check . I have seen ms malware scanning kick in on files accessed for copying.. and it takes awhile to scan inside of zips.. downloads folder sounds like one of its targets. Maybe check process wise with resource monitor and see what's doing disk accesses around the times you have the copy running.

 

[Fritz]

Moved all 6 drives back to a SM chassis and they all test good and all copy operations, including a downloads folder proceed normally. I know the MB is good because it was previously in a SM chassis and gave me no problems. I suspect a flakey PSU. Will know shortly. One of the drives suddenly lost it's name and reverted to "local Disk". Put the drive in another box and it reverted back to it's normal name so something isn't right here. I suspect the PSU. It's an older Thermaltake 750w.