landiag processing taking up most of my cpu resources

[James Ellis_1]

I noticed landiag process was taking up to 89% of three CPUs. We have 4 CPUs and this process is hogging 89% of each three CPUs. Is this normal? I want to verify if this is what usually happens. I'd like to turn it off and have most of the CPU resources available for developers.

The man pages say lanadmin replaces the now obsolete landiag command beginning at 10.0. I have 11.0 so the system should be using the lanadmin, but why is landiag still showing up in the process?

If I wanted to quit landiag, I would issue the quit command the same as you would quit the lanadmin?
Thanks.

[John Poff]

Hi,

I'm not sure why your 11.0 box is running landiag. As you mentioned it is replaced by lanadmin. Is there a tty associated with the landiag process? Maybe someone found the old utility and ran it?

My suggestion is to figure out who or what ran it, and why, and then just kill the process.

JP

[John Poff]

Hi again,

On one of my 11.00 systems, the landiag and lanadmin commands look like they are hard linked to the same executable:

# ls -ld /usr/sbin/landiag
-r-xr-xr-x 2 root bin 49152 Jul 20 2001 /usr/sbin/landiag

# ls -ld /usr/sbin/lanadmin
-r-xr-xr-x 2 root bin 49152 Jul 20 2001 /usr/sbin/lanadmin

Are they the same on your box or are they two different executables?

JP

[Helen French]

Check for a patch in your system - PHNE_26250 (s700_800 11.00 LAN product cumulative patch). If not found, try installing it. That may solve the issue.

[James Ellis_1]

They are both the same executables. I may check my patches and see if they are up to par. Thanks.

[James Ellis_1]

I got the patch already installed on my system. I tried to kill the landiag, and it just would not terminate. Do you kill it by going ot the command line and type landiag and then type quit?

[John Poff]

Hi James,

I ran landiag here and it looks like the same thing as lanadmin.

Can you tell if your landiag is running with any parameters, or is it running standalone? If it runs without any command line parameters it should come up in the menu interface. Since you can't kill the process I would suspect that it was trying to change a LAN card setting or query something and got hung up on some I/O. Ouch. Are you getting any complaints in syslog or maybe from nettl?

JP

[James Ellis_1]

John,
Yeah about 2 weeks ago I installed a new card (A4926A = fiber gig card), and unplumbed the first card and set up the second card. Card seem to work but I just noticed that the landiag is using up way too much resources.

Oh, I tried to kill landiag by running landiag and then quit. I guess I could kill the process, but I initially thought my first way would be more clean?

Thanks.

[Sridhar Bhaskarla]

Hi James,

landiag is not a process that runs all the time like a daemon. As you already know, it is used only during diagnostics. Are you running any script that runs landiag continuously?. I believe not.

When you invoke another landiag, it is going to be another process altogether. So, exiting it cleanly will not affect the one that was already running.

It may be an orphaned process. Go ahead and kill it.

-Sri

[James Ellis_1]

Sridhar,

It looks like it keeps popping up or starting up. I am not sure what causes it to start up, but this morning I two killed landiags that were running. And about 30 minutes later it appears another process has started but this time its only one landiag running.

I never had this problem until I installed the gelan cards on two of my servers. Both servers show landiag running (either 1 or 2 or 3 landiag processes) at one time or another, and often they are taking up 99% of a cpu, if not more than 1.

It appears something is starting up landiag and I am going to check the patch level and requirement. Surely this has something to do with the gelan driver/NIC?

Thanks.

[John Poff]

Hi,

What is the PPID of the landiag process? Maybe that will give some clues. My guess is that some kind of automatic diagnostics are running on the box. Could EMS be using landiag to check on the health of the LAN cards?

JP

[James Ellis_1]

John,

I think its some kind of automatic diagnosis running on the box, but am still looking to find where/what is the purpose for this. By the way, what is PPID, or rather, how can you find the PPID?

Thanks.

[John Poff]

Hi,

The PPID is the parent process ID number. Like this:

# ps -fp 23008
UID PID PPID C STIME TTY TIME COMMAND
root 23008 2742 0 Jan 25 ? 0:22 diaglogd

The parent PID of my diaglogd daemon is 2742, so I can go look at that:


# ps -fp 2742
UID PID PPID C STIME TTY TIME COMMAND
root 2742 1 0 Jan 25 ? 1:23 diagmond


...and see that the parent process is the diagmond daemon.


If you can figure out what the parent process is of your landiag processes, you might have a good clue about what is spawning them.

JP

[Sridhar Bhaskarla]

HI James,

I believe your landiag has been getting invoked by scripts either through cronjobs or your monitoring softwares like VPO, Tivoli etc., whichever has been configured on your system.

Take a ps -ef output and note down the ppid of the landiag process as mentioned by John. Search for ppid in the same output and see if you can find a parent process. I think HAO does invoke landiag and you may want to check it too if you have it.

-Sri

[James Ellis_1]

Ok, this is what I got out of the system. I tried to find the parent process:

ps -fp 2182
PID=2182, PPID=2181 Time=6735 Command=landiag -t

ps -fp 2181
PID=2181, PPID=1, Time=0 Command= sh -c landiag -t 2>/dev/null <
As you can see, the parent process is the command that ran landiag. I don't see anyplace where this command is ran. Not only that, it appears there is always TWO running at the same time.
Thanks.

[Dietmar Konermann]

First of all, you should install a patch for your interface card. Although landiag is onsoleted, it should not hog a CPU. This should be fixed with PHNE_25612, if gelan is the driver you are using.

Looking at the arguments of landiag I strongly suspect the Unicenter TNG Agent to do this kind of calls. Would this be possible?

Best regards...
Dietmar.

[Sridhar Bhaskarla]

Hi James,

The pid 2181 is a sub shell running landiag. Since it's parent is 1, it could most probably that it is starting through cron. Look at your crontabs.

Try killing the pid 2181 and see if it respawns.

-Sri

[James Ellis_1]

I recall killing pid 2181, and it respawned. But I'll try again. for some strange reason, landiag seems to respawn and have 3 landiag processes running at the same time. I checked the crontab files and it does not look like any cron job is opening a subshell.

I will apply the PHNE_25612 patch and see if this helps. Yes, we are running Unicenter, and I asked the Uni Admin if he knew of any jobs that would start up lanadmin, and he said as far as he knows he does not think so.

I'll get the patch installed and let you all know how it goes. Thanks.