topic Re: Serious problem with VA7400 in Disk Enclosures

Serious problem with VA7400

Domenico Viggiani — Mon, 28 Jan 2002 17:50:46 GMT

I have a VA7400 disk-array connected by a Brocade 2800 switch to a RP7450 machine with HP-UX 11.00 64 bit.
Two times, during restore of 50-70 GB of data from an Ultrium tape to VA7400 disks, machine hung and I needed to hit power button!

OLDsyslog.log is full of these messages:

Jan 28 18:43:45 prod1 vmunix: SCSI: Async write error -- dev: b 31 0x110500, errno: 126, resid: 32768,
Jan 28 18:43:45 prod1 vmunix: blkno: 72993600, sectno: 145987200, offset: 1731002368, bcount: 32768.
Jan 28 18:43:46 prod1 vmunix: SCSI: Async write error -- dev: b 31 0x110500, errno: 126, resid: 24576,
Jan 28 18:43:46 prod1 vmunix: blkno: 72993536, sectno: 1...
Jan 28 18:43:46 prod1 vmunix: LVM: Recovered Path (device 0x1f110500) to PV 0 in VG 5.
Jan 28 18:43:46 prod1 vmunix: LVM: Performed a switch for Lun ID = 0 (pv = 0x0000000042b7a800), from raw device 0x1f130500 (with priority: 1, and current flags: 0x40) to raw device 0x1f110500 (with priority: 0, and current flags: 0x0).
Jan 28 18:43:46 prod1 vmunix:
Jan 28 18:43:46 prod1 above message repeats 221 times
Jan 28 18:43:46 prod1 vmunix: LVM: vg[5]: pvnum=0 (dev_t=0x1f110500) is POWERFAILED
Jan 28 18:43:46 prod1 vmunix: LVM: Restored PV 0 to VG 5.
Jan 28 18:43:46 prod1 vmunix: LVM: Recovered Path (device 0x1f130500) to PV 0 in VG 5.

Last week, a LUN from the same array disappeared.

It seems a serious issue.
I'm waiting for HP support but I'd like to hear a suggestion form you.

Thanks in advance
Domenico Viggiani

Re: Serious problem with VA7400

Robert_Jewell — Tue, 29 Jan 2002 18:28:42 GMT

It looks like you have drive going on you. From the message:
LVM: vg[5]: pvnum=0 (dev_t=0x1f110500) is POWERFAILED
it shows that the device on c10t5d0 is giving you the problem. If you run an # ioscan -fnkC disk you should see what physical device it is referring to. You could post the results from ioscan if you like.

-Bob

Re: Serious problem with VA7400

Domenico Viggiani — Wed, 30 Jan 2002 09:31:40 GMT

'ioscan -fn' hangs forever.
c10t5d0 is a DLT drive and it is unrelated to this problem. I think.

Re: Serious problem with VA7400

Robert_Jewell — Wed, 30 Jan 2002 23:50:21 GMT

I apologize, I appear to have shifted the numbers a bit:

1f11050

1f coverted to dec. = 31 which is major # for sdisk.
11 = bus #
0 = target #
5 = LUN

so...c11t0d5. Does that make more sense??

-Bob

Re: Serious problem with VA7400

Domenico Viggiani — Thu, 31 Jan 2002 08:25:32 GMT

No. There is no c11t0d5.

Thanks anyway.

Re: Serious problem with VA7400

Stephan_K — Thu, 31 Jan 2002 12:00:34 GMT

Hi,

we had similar problems.
During Backup the VA hanged up and also with the armdsp command we did receive die VA7400. We needed to remove one controller from the va and put it back into.
HP support presented us a workaround which was definining the Resilency Mode "HighPerformance" for the VA. But it solved not really the problem. The solution was to make an FirmwareUpgrade of the VA to HP14.
But we were told that with the next fw version are again some performance issues. So we will also install this fw.
But with HP14 and "HighPerformance" mode the VA runs ok.
Please take care of the reduced data protection you have with "highPerformance" mode.
Regards
Stephan

Re: Serious problem with VA7400

Domenico Viggiani — Thu, 31 Jan 2002 18:39:12 GMT

Actually, I have HP13 fw revision.
Array log is filled with following messages (since days of problems):

FRU Location = M/C1.H1
Event Number ---------------- = 17,602
Sequence Number ------------- = 1,269
Time ------------------------ = Jan 31, 2002 7:35 PM
Device Name ----------------- = HP SureStore Virtual Array 7400
Device Unique Name ---------- = 50060b000008aa33
Device Type ----------------- = DiskArray
Event Code ------------------ = 348
Event Description = FRONTEND_FC_ABTS_EVENT_EH This error code indicates that the Host sent
a Fibre Channel ABTS (Abort Sequence) BLS frame to the abort an IO. The array will log thi
s event for informational and debug purposes only. It does not necessarily indicate a prob
lem with the array.
Event Unique Data :
Ckpt 0x470097: 0x00000004
frontend: osIOCompleted: osIOAbortABTS
NPortID=0x011000,OX_ID=0x00fe,LUN=0x4001
CDB=0x2a000566e6a000001000000000000000
ElapsedTime=31.1 ioHWM=0x0023 jstate=0x05
controller tick : 1492421663479
serialNum/moduleId/processId : 00PR0301V566/0x47/0xffffffff
enclosureId/slot : 0x00/0x00

Re: Serious problem with VA7400

Jeroen Ruigrok vd Werve_2 — Wed, 20 Feb 2002 10:43:14 GMT

Mmm,

the dev_t is hexadecimal all over.

So the 11 would also need to be converted to decimal, in this case would be c17t0d5.

Re: Serious problem with VA7400

Domenico Viggiani — Thu, 21 Feb 2002 08:44:11 GMT

Problem was solved by HP support configurung the array for 'maximum performance', as suggested in a previous message and according to a HP doc.

Thanks to all

Re: Serious problem with VA7400

Saurav_1 — Thu, 19 May 2005 07:24:37 GMT

Hello Domenico,

We have a same problem with VA7410. We also face lun switch problem in our servers. Can you tell me in detail how this problem was resolved and exactly which setting resolved this issue.

Regards

Re: Serious problem with VA7400

Domenico Viggiani — Thu, 19 May 2005 07:36:26 GMT

Saurav,
too time is gone and I don't remember anymore!
Anyway, I think that, as said above:

"with HP14 and "HighPerformance" mode the VA runs ok.
Please take care of the reduced data protection you have with "highPerformance" mode."

solves the problem.

I still have that VA7400 running without issues.