nmon problem 2

[sbrews]

While trying to run nmon, I encounter the following error:

no AIX version (can't run lslpp)
nmon file=nmon9.c line=2270 version=v9a

Nmon used to work without issue. It just started doing the above recently. As far as I know, nothing on the system has changed (but other admins have acces too, so... (but they all claim to not have made any changes))

Various checks (lppchk, instfix, ...) show that everything is as expected.

Has anyone run into this particular error? If so, how did you fix it?

System specifics:

nmon v9a
LPAR running AIX 5.2 ML3
6g memory
3 cpu
all mount points have plenty of free space

scott

 

[RodKnowlton]

Could be a PATH problem, or maybe permissions on /usr/bin/lslpp have been changed.

nmon appears to run (looking through strings output) the following to find out your AIX version:

lslpp -Lcq bos.?p 2>/dev/null  |tail -1 |cut -f3 -d:

I'd take it at it's word (can't run lslpp), unless you've already tested that you can run lslpp with the same user and environment with which you're trying to run nmon.

Rod Knowlton
IBM Certified Advanced Technical Expert pSeries and AIX 5L
CompTIA Linux+
CompTIA Security+

 

[sbrews]

I have checked permissions, etc., all are as they should be. Further testing shows that I can issue this command without issue:

lslpp -Lcq bos.?p 2>/dev/null

However as soon as I try to pipe any other commands (doesnt matter which command), I get a "Segmentation fault(coredump)".

The above core only appears to happen with the lslpp command. Other commands, such as ls or ps do not appear to have a problem with pipes (yet).

Anyone have any ideas why lslpp cores when I pipe it?

scott

 

[sbrews]

minor update:

While lslpp -Lcq bos.?p does produce output, it also core dumps.

 

[sbrews]

minor update #2:

(gotta remember to check everything before posting updates)

The above listed lslpp command works just fine if I dont use the ? wildcard.

 

[RodKnowlton]

Take a look at

lslpp -h bos.rte.install

and see if there's been a patch whose date corresponds with the sudden change.

You may also want to do

lppchk -c bos.rte.install /usr/bin/lslpp

to make sure the binary's not corrupted.


Rod Knowlton
IBM Certified Advanced Technical Expert pSeries and AIX 5L
CompTIA Linux+
CompTIA Security+

 

[sbrews]

Blah! Everything checks out. No patches, no corruption.

Any other ideas/areas to check would be greatly appreciated.

scott

 

[sbrews]

ok, interesting development.

I did a "bosboot -ad /dev/hdisk0". The lslpp command now works correctly interactively. Nmon, however, still fails. I will get the box rebooted and try again.

Details once the reboot is done.

Thanks to all who have responded.

scott

 

[mnbhahn]

scott,

I'm having the same problem with nmon (lslpp - segmentation fault only when I use the "?" wildcard). Example...

# lslpp -Lcq bos.?p
bos.mp:bos.mp:5.2.0.41: : :C:F:Base Operating System
bos.up:bos.up:5.2.0.41: : :C:F:Base Operating System
Segmentation fault
# echo $?
139

# lslpp -Lcq bos.mp
bos.mp:bos.mp:5.2.0.41: : :C:F:Base Operating System
# echo $?
0

- Mike

 

[RodKnowlton]

You could create a dummy lslpp, outside of your regular PATH, that gives the output that the real one should be giving to nmon, then launch nmon from a script that puts the dummy lslpp at the head of the PATH.

Of course, this will fail miserably if nmon uses lslpp for any other purpose.

I guess a solution in that case would be for the dummy lslpp to pass all of its arguments to the real lslpp, then strip off the Segmentation fault message and return true. I wouldn't do this first, though, since a dummy that always returns "-Lcq bos.?P" output will be harder to confuse with the real thing.



Rod Knowlton
IBM Certified Advanced Technical Expert pSeries and AIX 5L
CompTIA Linux+
CompTIA Security+

 

[lfrisvold]

We recently started having this problem also. We discovered that if we cd into /lpp or /usr/lpp, nmon can be ran without any trouble. Still trying to figure out why. Is this the same for you?

 

[sbrews]

Our problem turned out to be a bug in AIX 5.2 ML3. I do have another system at the same level that is not showing any problems. The bug didnt show up until someone installed Java 1.31.

After email with Nigel (nmon creator), the suggested fix (until a patch kit for the bug is available) is to set and export an environment variable:

export NMONAIX=5.2.0.0

Once we did that, nmon resumed working as expected.

scott

 

[mnbhahn]

I'm running 5.2 MR4. I don't think my problem was with NMON. I must have installed something and didn't "bosboot/reboot".

This is what I did to fix the problem...

# cd /usr/lpp
# nmon
nmon read error: No such device or address
nmon file=nmon9.c line=1736 version=v9a
nmon 99% of the time this is due to
- not rebooting AIX after and upgrade/fix and told to do so by installp
- this means /unix does not match the running AIX kernel
- check by seeing of netpmon or vmstat work correctly

# netpmon

Enter the "trcstop" command to complete netpmon processing

/unix does not match the executing kernel.
Exiting, as netpmon can not obtain valid kernel addresses.

Called IBM..

1) Applied APAR (IY59427)
2) bosboot (both hdisk0 & hdisk1)
3) rebooted

Problem solved.