GlancePlus won't work

[Dean Irving]

All...
Need help. GlancePlus quit working on our servers. Here's error msgs. Any ideas??

GlancePlus fatal error.

Tue Feb 17 10:38:32 2004
mi_shared - MI initialization failed (status 5)

== Fatal Nums Error == C.02.65.00 09/22/00 ==
User: irvingd(/dev/pts/ti) Date: Tue Feb 17 10:38:32
File: nums.C Line: 288 Product id: Glance
System: gusdmm01 B.11.11 9000/800
Errno: 2 (No such file or directory)

Midaemon not responding [MI_SHARED:] -- Error Text follows

Not enough space.
midaemon: Thu Feb 12 10:27:43 2004
mi_create - Shared Memory Database initialization failed; vss = 43019900 bytes
Not enough space.
midaemon: Thu Feb 12 10:28:33 2004
mi_create - Shared Memory Database initialization failed; vss = 43019900 bytes
Not enough space.
midaemon: Mon Feb 16 00:12:48 2004
mi_create - Shared Memory Database initialization failed; vss = 43019900 bytes
Not enough space.
midaemon: Mon Feb 16 00:31:23 2004

[Patrick Wallek]

It appears that you are running short of Shared Memory. Is this on a server running 32-bit HP-UX? If so, you are probably bumping up against the 32-bit ceiling for shared memory.

What else is running on this machine?

[Dean Irving]

Server is 16CPU SuperDome with 32GB memory running 11i Here's top output:
Memory: 5179308K (2865724K) real, 7201040K (3917216K) virtual, 20230160K free P
age# 1/121

[Patrick Wallek]

Have a look at this TKB document (Doc ID OV-EN013896)
http://www2.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000069748546

Basically the doc says to take a look at your SHMMAX kernel parameter. What is the setting? Has it been changed recently? It may need to be increased, which will require recompiling the kernel and a reboot.

[Dean Irving]

Patrick...
Thanks. Unfortunatel, SHMMAX already at 64MB. Here's kmtune output:
shmmax 6442450944 Y 6442450944

[Dave Hutton]

Actually thats the default. You can go up to 1 TB.

Minimum
2048 (2 Kbytes)
Maximum
0x40000000 (1024 Mbytes) on 32-bit systems
Maximum 1 Tbyte on 64-bit systems
Default
0x04000000 (64 Mbytes)

http://docs.hp.com/hpux/onlinedocs/939/KCParms/KCparam.ShMmax.html

So you can bump that up quite a bit.

[john korterman]

Hi,
may sound idiotic, but what is the total pct. swap used?
# swapinfo -t

regards,
John K.

[Dean Irving]

swapinfo -t
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4194304 0 4194304 0% 0 - 1 /dev/vg00/lvol2
dev 8388608 0 8388608 0% 0 - 1 /dev/vg00/lvol10
dev 8388608 0 8388608 0% 0 - 1 /dev/vg00/lvol11
dev 8388608 0 8388608 0% 0 - 1 /dev/vg00/lvol12
reserve - 6981980 -6981980
memory 26164016 4930696 21233320 19%
total 55524144 11912676 43611468 21% - 0 -

[Dean Irving]

Here are my shared memory kernal settings.
shmem 1 - 1
shmmax 6442450944 Y 6442450944
shmmni 400 - 400
shmseg 120 Y 120

[john korterman]

Hi,
it was an idiotic suggestion; absolutely no swap problem.

Please try to check the number of active semaphores:
# ipcs -s | wc -l
is that figure close to the limit, as shown by:
# echo "shmmni/D" | adb -k /stand/vmunix /dev/kmem

regards,
John K.

[Dean Irving]

Here's output...doesn't look like problem

$echo "shmmni/D"|adb - k /stand/vmunix /dev/kmem
shmmni:
shmmni: 400
$ ipcs -s |wc -l
30

[Elena Leontieva]

Hi,

Try to replace the /etc/magic file with a copy from /usr/newconfig/etc/magic

Elena.

[Sridhar Bhaskarla]

Hi,

Whatelse you have on the box?. I don't think it is related to shmmax parameter as glance would never start on the box.

This is a 32-bit process and you are limited to a maximum of 1.75GB of shared space unless you are using memory windows for anyo f your 32-bit applications. What is your SGA size if you are running oracle and what other applications are running on the box?.

Try running it with 'tusc' and see|attach the output. You would know where it is failing.

tusc -o /tmp/glance.out glance

-Sri

[john korterman]

Hi again,
what has happened on the system(s) since the last successful start of GlancePlus? Has anybody made any experiments: chmod or the like?

regards,
John K.