My thought process was going down a similar line as Scott's. You may have
too many busy logical disks configured on a single spindle. EMC can run
some collection software to analyze the box if you ask. There may be a
charge for this service though.
You could also purchase EMC's Symm Manager, similar to Glance, and Workload
Analyzer, a collection tool/analyzer like Scope, so that you can monitor the
Symm any time you want. They are a bit pricy but if you hold your own
against the sales rep you can get the price down. Just remember that with
EMC everything is negotiable.
On a similar note, I managed to crash one of our systems a while back by
accidentally configuring all the logical disks from a single spindle into
one busy volume set. A test program was doing lots of disk writes which
filled the EMC's cache and then backed up the HP's I/O Buffer table and
BOOM! You can check for this type of problem on the MPE side by running
THMGR and watching the really neat charts that it displays. At the time I
didn't think to look at the logs to see if any hardware errors were logged
though.
Carl McNamee
Systems Administrator
Billing Concepts
(210) 949-7282
-----Original Message-----
From: Scott Lindholm [mailto:[log in to unmask]]
Sent: Tuesday, July 18, 2000 5:28 PM
To: [log in to unmask]
Subject: Re: [HP3000-L] EMC disc timeouts
Budd,
We just had a meeting with our EMC ASE and went over some performance data.
Our ASE showed us a metric called Write Pending Limits. As he explained it,
if we reached the limit the Symm would tell the 3000 that it was busy and
the 3000 would log an error. I do not know if this is your problem but you
might want EMC to take a peak inside and see what is going on.
Good Luck,
Scott Lindholm
Sr. Systems Administrator
1 Environmental Way
Broomfield, Colorado 80021-3416
(303) 664-3264
[log in to unmask]
> -----Original Message-----
> From: Budd Marsh [SMTP:[log in to unmask]]
> Sent: Tuesday, July 18, 2000 3:07 PM
> To: [log in to unmask]
> Subject: [HP3000-L] EMC disc timeouts
>
> I am logging timeouts for discs in an EMC Symmetrix box. The timeouts are
> happening on discs on path 8/12/36 which is configured as a(hp28696a)
> FXSCSI
> DAM. The errors are showing up as Type 111 I/O errors in the logfiles. I
> am getting about 15-20 timeouts daily on the discs on this path and also
> seeing about 5-10 I/O errors logged for the 28696A daily.
>
> HP has replaced the 28696a card and I still have the errors. I have also
> had the SCSI cable to the EMC box replaced. This card is in a Bus
> Expansion
> unit attached to the HSC card in my 979-300. We are planning to replace
> the
> incoming card on the Bus Expansion unit tonight. Has anyone else seen
> similar errors? Can anyone explain what happens to the transaction that
> timed-out? I am having frequent out-of-sync conditions with databases
> being
> shadowed using Netbase because the pre-update/pre-delete image on this
> machine does not match the master.
>
>
> ==========================================================================
> ==
> ===
> MON, JUL 17, 2000 7:27 AM LOG0141.PUB.SYS SYSTEM
> (PIN
> 83)
> I/O ERROR
> PRODUCT NAME: SYMMETRIX PDEV: 8/12/36.12.0
> LDEV: 82 DEVICE CLASS: Hard Disc
> I/O EVENT CLASS: Hardware LLIO STATUS: $CFCC0384
>
> A TIMEOUT has occurred, indicating that the drive is not talking to the
> device manager.
>
> RETRY SCHEME: Log Each Retry WILL RETRY: NO
> I/O RESULT: I/O Failed RUN AUTODIAG: NO
> RETRY COUNT: 0 MGR PORT NUM.: $FFFFFF35
> TRANS. NUM.: $D2C00550 # HDWR BYTES: 36
> HARDWARE STATUS:
>
>
> Sense Key Code = 0 ($00)
> No Sense Key Information
> Additional Sense Code (ASC) = 0 ($00)
> Add'l Sense Code Qualifier (ASCQ) = 0 ($00)
> No additional sense information. No error.
>
> Note: The PDEV/LDEV data shown above is only a guideline.
> For precise identification of the unit, use the following information:
> Cabinet S/N: (last 6 digits)
> Physical Volume: (EMC device number)
> SA Number:
> SA Port Number:
>
> Raw Sense Bytes (in hex):
> 0 3 4 7 8 11 12 15 16 19 20 23 24 27
> 29
> -------- -------- -------- -------- -------- -------- --------
> ----
> 00000000 00000000 00000000 00000000 00000000 00000000 00000000
> 0000
>
> DATA LEN: 28 MGR CODE: 900
> SCSI CMD BUFFER:
> 1 2 3 4 5 6 7 8 9 10 11 12
> == == == == == == == == == == == ==
> 1: 28 00 00 5F E8 20 00 00 80 00 00 00 ( . . _ . . . . .
> .
> .
>
> MSG(bytes 0-15):
> 1 2 3 4 5 6 7 8 9 10 11 12
> == == == == == == == == == == == ==
> 1: 02 6D 00 0B D2 C0 05 50 FF FF FF 3E . m . . . . . P . .
> .
> >
> 13: D7 50 03 85 . P . .
>
>
> ==========================================================================
> ==
> ===
> MON, JUL 17, 2000 7:41 AM LOG0141.PUB.SYS SYSTEM
> (PIN
> 83)
> I/O ERROR
> PRODUCT NAME: HP28696A PDEV: 8/12/36
> LDEV: DEVICE CLASS: DA
> I/O EVENT CLASS: Software LLIO STATUS: $A1AB0385
>
> MPE/XL I/O Status: Proc. Num. = -85, Error Num. = -95, Subsystem = 901
>
> RETRY SCHEME: Summarized Retries WILL RETRY: NO
> I/O RESULT: I/O Failed RUN AUTODIAG: NO
> RETRY COUNT: 0 MGR PORT NUM.: $FFFFFF3E
> TRANS. NUM.: $0 # HDWR BYTES: 0
> HARDWARE STATUS:
>
> No hardware status was logged.
>
> DATA LEN: 36 MGR CODE: 901
> FWSCSI Error : $A1 FWSCSI Proc. : $AB
> FWSCSI Subsys. : $385 LoQuix IO_Stat : $1
> Wizard Int_Stat: $601F Wizard U_Event : $8800080
> Wizard Ctrl_Req: $FFFFFFF0 Wizard Tst_Sw : $1
> Wizard FW Mode : $1 Wizard PassThru: $1
> Wizard TermPWR : $1 Wizard TP_Sense: $1
> Wizard In_Bank : $0 Wizard P_Bank : $0
> Wizard A_Bank : $0 Wizard Parity : $1
> Wizard Address : $7 Wizard FW DCode: $3728
> SPIFI Int_State: $F SPIFI Cur_State: $F
> SPIFI Proc_Intr: $1 SPIFI Proc_Stat: $FF
> SPIFI Intr_Cond: $F0 SPIFI Init_Stat: $FF
> ==========================================================================
> ==
> ===
>
> Budd Marsh
> System Administrator
> Multiple Zones Int'l
> [log in to unmask]
Billing Concepts Corp., a NASDAQ listed Company, Symbol:BILL
|
