Disk problems running Sysplex under VM

[s***@virgin.net]

I'm running z/VM Version 4 Release 3.0, service level 0201 on a
multiprise 3000 (7060)
I'm setting up a Sysplex using 3 Z/os 1.5 but I keep getting
"HCPVER575I I/O error add=0100, userid= ZPLEX2" on shared disks
causing the systems to hang.
I have WRKALL set on the minidisk that holds the control data sets and
SHARED on all the full packs. There are no hits on IBM problem database
has anyone come across this problem.

Regards

Stuart.

[Rob van der Heij]

Post by s***@virgin.net
I'm setting up a Sysplex using 3 Z/os 1.5 but I keep getting
"HCPVER575I I/O error add=0100, userid= ZPLEX2" on shared disks
causing the systems to hang.

So what *is* the error? Depending on your configuration they will be
recorded in EREP on z/OS or z/VM. Alternatively, you could run an I/O
trace to see what happens.

If the MP3K is the only physical machine involved (and I believe it
will since I don't think you can mix virtual machines and LPARs in a
sysplex) then the SHARED is not needed. You will have your shared mini
disks defined on one guest as MWV, and have the others link MW to it,
right?

Rob
--
Rob van der Heij
Velocity Software, Inc

[s***@virgin.net]

Hi Rob,

Yes the MP3K is a single physical machine.

Long time since I used EREP I will see what I can remember ;-)

The error I get in Zos system is
$HASP9201 JES2 MAIN TASK WAIT DETECTED AT ISGGWAIT+00014C 564

DURATION-000:00:12.93 PCE-CKPT EXIT-NONE JOB ID-NONE

IEF196I IOS078I 0A84,FD,XCFAS, I/O TIMEOUT INTERVAL HAS BEEN EXCEEDED

IEF196I FOR AN ACTIVE REQUEST. THE ACTIVE REQUEST HAS BEEN

IEF196I TERMINATED.

IEF196I QUEUED REQUESTS MAY HAVE ALSO BEEN TERMINATED.

IOS078I 0A84,FD,XCFAS, I/O TIMEOUT INTERVAL HAS BEEN EXCEEDED 565

FOR AN ACTIVE REQUEST. THE ACTIVE REQUEST HAS BEEN TERMINATED.

QUEUED REQUESTS MAY HAVE ALSO BEEN TERMINATED.

IEF196I IOS071I 0A84,**,OMVS, START PENDING

IOS071I 0A84,**,OMVS, START PENDING 570

IEF196I IOS079I 0A84,FD,XCFAS, I/O TIMEOUT INTERVAL HAS BEEN EXCEEDED

IEF196I FOR A QUEUED REQUEST. THE QUEUED REQUEST HAS BEEN

IEF196I TERMINATED.

Stuart.

[Rob van der Heij]

Post by s***@virgin.net
Long time since I used EREP I will see what I can remember ;-)

No doubt more than what I know about the z/OS msgs...

Maybe an I/O trace is at least as easy. In the z/OS virtual machine
(where xxx is the virtual address to trace)
#cp trace io xxx ccw printer
When you are done you stop the trace and send the print file to MAINT
so you can see it.
#cp trace end
#cp sp prt maint close

I expect z/VM is just simulating the I/O error for z/OS.

--
Rob van der Heij
Velocity Software, Inc

[s***@virgin.net]

Rob

O/P from erep.

DEVICE NUMBER: 0100 REPORT: MIH EDIT DAY YEAR
JOB IDENTITY: ZPLEX2
SCP: VS 2 REL. 3 DATE: 067 06
E9D7D3C5E7F24040
DEVICE TYPE: 3390

CPU MODEL: 7060 HH MM
SS.TH
CHANNEL PATH ID: N/A CPU ID: 010BD9 TIME: 13 00
04.10
MISSING INTERRUPT: 08 - I/O TIMEOUT CONDITION SUBCHANNEL ID
NUMBER: 00010002
FOR AN ACTIVE REQUEST VOLUME SERIAL:
ZVMRES
HH MM SS.TH UCB LEVEL
BYTE: 01
TIME INTERVAL: 00 00 15.00

RECOVERY ACTIONS PERFORMED BYTE: CC

HALT OR CLEAR SUBCHANNEL 1

SIMULATED INTERRUPT 1

REDRIVE DEVICE 0

REQUEUE I/O REQUEST 0

ISSUE MESSAGE 1

LOG THE CONDITION 1

BIT 6 0

BIT 7 0

HEX DUMP OF SUBCHANNEL INFORMATION BLOCK

OFFSET 00F3D7C8 289B0A84 80008080 0127FF80

0010 FDFFFFFF FFFFFFFF 00000001 03C04400

0020 00000000 00000000 00000000 00000000

0030 00000000



HEX DUMP OF RECORD

HEADER 71831800 00000000 0006067F 13000410 1B010BD9 70600000

0018 E9D7D3C5 E7F24040 00F3D7C8 289B0A84 80008080 0127FF80
FDFFFFFF FFFFFFFF
0038 00000001 03C04400 00000000 00000000 00000000 00000000
00000000 F0F0F0F0
0058 F1F5F0F0 08CCCCCC 00010002 28988080 80FDFFFF FFFFFFFF
FF010000 00000100
0078 00000100 08008005 2024E9E5 D4D9C5E2 00008000 09000063
FF0001FF 00001001
0098 00000000 00000000 00000000



DEVICE NUMBER: 0100 REPORT: MIH EDIT DAY YEAR
JOB IDENTITY: ZPLEX2
SCP: VS 2 REL. 3 DATE: 067 06
E9D7D3C5E7F24040
DEVICE TYPE: 3390

CPU MODEL: 7060 HH MM
SS.TH
CHANNEL PATH ID: N/A CPU ID: 010BD9 TIME: 13 00
20.08
MISSING INTERRUPT: 10 - START PENDING IN SUBCHANNEL SUBCHANNEL ID
NUMBER: 00010002
VOLUME SERIAL:
ZVMRES
HH MM SS.TH UCB LEVEL
BYTE: 01
TIME INTERVAL: 00 00 15.00

RECOVERY ACTIONS PERFORMED BYTE: AC

HALT OR CLEAR SUBCHANNEL 1

SIMULATED INTERRUPT 0

REDRIVE DEVICE 1

REQUEUE I/O REQUEST 0

ISSUE MESSAGE 1

LOG THE CONDITION 1

BIT 6 0

BIT 7 0

HEX DUMP OF SUBCHANNEL INFORMATION BLOCK

OFFSET 00F3D7C8 289B0A84 80008080 0127FF80

0010 FDFFFFFF FFFFFFFF 00000001 03C04400

0020 00000000 00000000 00000000 00000000

0030 00000000



HEX DUMP OF RECORD

HEADER 71831800 00000000 0006067F 13002008 1B010BD9 70600000

0018 E9D7D3C5 E7F24040 00F3D7C8 289B0A84 80008080 0127FF80
FDFFFFFF FFFFFFFF
0038 00000001 03C04400 00000000 00000000 00000000 00000000
00000000 F0F0F0F0
0058 F1F5F0F0 10BCBCAC 00010002 28988080 80FDFFFF FFFFFFFF
FF010000 00000100
0078 00000100 08008005 2024E9E5 D4D9C5E2 00008000 01000063
00000100 00001001
0098 01C06401 00000000 00000000


Stuart

[Rob van der Heij]

So that's MVS' missing interrupt handler kick in? Looks like CP is not
reflecting anything on this... maybe the I/O trace would reveal
what's going on.

Rob
--
Rob van der Heij
Velocity Software, Inc

[s***@virgin.net]

Rob I have the Trace all 122252 lines of it what should I be looking
for as it's not something I have looked at before.

Stuart.

[s***@virgin.net]

Solved the problem. Looking at the Trace ETC I noticed that ZOS was
issuing the "JES WAIT DETECTED" message before VM was issuing the
"HCPVER575I I/O error" therefore it was ZOS crating the problem. To cut
a long problem short you have to EXCLUDE the JES checkpoint data sets
from GRS as per the Manual :-(
All ok now thanks to Rob for his help.

Stuart