System load is very high (Load averages: 15.28, 15.27, 15.27) and more sleeping process (217)

[Raj D.]

Senthil,

> How to kill those orphaned shells?


root 43 0 0 Oct 10 ? 1:09 vx_attrsync_thread
root 44 0 0 Oct 10 ? 1:03 vx_inactive_thread

- You have to reboot the box to clear these inactive processes, nothing can be done online. It seems to be server has not been rebooted for long time 212days+ from your earlier output.

- Seems to be your system need patching. What is the current patch level of this 10.20 box.

- It is also recomended to disable online JFS drivers if not using: SAM->kernel->drivers:
dmapi
vxbase
vxadv


- You can check patch level using:
# swlist -l bundle | grep -i -e QPK -e GOLD -e HWE -e Online -e HPUX11

further check out this link:
http://forums13.itrc.hp.com/service/forums/questionanswer.do?admit=109447627+1277247325525+28353475&threadId=712678


Hth,
Raj.

[chris huys_4]

Hi,

> - It is also recomended to disable online
> JFS drivers if not using:
> SAM->kernel->drivers:
> dmapi
> vxbase
> vxadv
No, its not recommended to disable these kernel drivers.

Never try to micromanage the OS. If the OS defaults to have these drivers enabled, then let the drivers be enabled and having no vx_inactive_threads, if they have allready anything to do with the enabled drivers, is certainly not a good enough reason for disabling the drivers.

Greetz,
Chris

[Raj D.]

Well, That was a point from HP , in the above links link.
Not sure how much effective that would be . However patches also needs to be check.

[Dennis Handly]

>1)what is the purpose of above command? ps -fp 17305

It should only list one PID. You seemed to be doing "ps -ef -fp 17305", which lists everything.

>2)How to kill those orphaned shells?

Unless you can figure out what device they are hung on, you can't.

[senthil_kumar_1]

Hi All,

Today onwards I am getting below error message while try to rlogin to this server.

Example:

#login serverA
rlogind: Unable to allocate pty on remote host.
Connection closed.


My Question:

1)The above issue is happening only because of lot of shells not closed properly?

2)If yes, How to terminate the shell automatically immediately after user exit from shell? I would like thie issue not to happen in future.

[Dennis Handly]

>1)The above issue is happening only because of lot of shells not closed properly?

Yes. Can you check who owns these /dev/tty or /dev/pty?

>2)How to terminate the shell automatically immediately after user exit from shell?

This should happen automatically. Can you duplicate this problem? Is some script trapping SIGHUP?

[Kranti Mahmud]

Hi Senthil,

Increase the kernel parameter nsrpty using SAM may resolve it.

Rgds-Kranti

[Kranti Mahmud]

Hi Again,

Check the below thread:

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=32314

Also the below link:
http://docs.hp.com/en/B2355-90692/rlogind.1M.html

Rgds-Kranti

[senthil_kumar_1]

Hi Dennis Handly,

/dev/pty and /dev/tty are owned by bin.

Example:

#ll -d /dev/pty
drwxr-xr-x 2 bin bin 8192 Feb 3 2005 /dev/pty

#ll -d /dev/tty
crw-rw-rw- 1 bin bin 207 0x000000 Mar 25 09:22 /dev/tty


Hi All,

I have checked value of following kernel parameters in SAM

Name Current_values Pending_values

npty 200 200
nstrpty 200 200


But in the output of sysdef it is showing like



#sysdef | grep -i npty
npty 60 - 1- -

#sysdef | grep -i nstrpty
nstrpty 60 - 1- -


My Questions:

1)Why the values shown in SAM and Sysdef differs? how to resolve?

2)Will the following commands resolve the issue?

# cd /dev
# insf -n200 -dpty0
# insf -n200 -dpty1
# insf -d pts -s 200 -e


3)Do I need to restart the server after execting above commands to take effect?

4)I am in the position of not restarting the server to resolve the issue, So is there any way to resolve the issue with out restart?

[Steven E. Protter]

Shalom,

My opinion on this saga.

A system restart is needed to make the kernel values:

npty and nstrpty

equal 200.

This will run insf -e as part of the startup.

If these kernel values are not dynamic, then a restart is your only option. That is definitely the case with 11.11

I'm in a new place and don't have an 11.31 system to check the kernel variables.

SEP