server unexpectedly closed network connection on live server

[subodhbagade]

To All HP-ux Expert,

I am facing one serious problem with hp-ux server 11.23, PA-RISC system having sap application on it. Need your valuable suggestion for the following scenario:----

(1) We are getting the error of “server unexpectedly closed network connection” while login remote server through putty (But I am able to connect the other server on same N/W using same putty.) Also users are not able to access the sap application.

(2)Also we are getting the Monitoring alert as “SERVER - ssh DOWN” for same server.
But one interesting thing is SERVER is PINGING but we are not able login the server.

(3) I have tried with GSP prompt and I am getting the GSP prompt but

After selecting “CO” option it is not giving me CON SOLE LOGIN.

(4)This is 3rd time we are facing same issue, First time we have Hard Rebooted the server physically.

2nd time I have reset the RST signal using “RS”.
After Rebooting server in both of the case I am able to login the server using putty (ssh).

(5) But this is 3Rd time in row we are facing the same problem, I have checked the syslog but unable to find the cause of this.

(6) As it is Live server, users are getting affected so Kindly suggest to rectify the issue. ( and let me know if any additional information, if require.)

Thank you Kindly, Have a Good time !!!!

Regards,
Subodh

[sen_ux · Accepted solution]

If sshd is down you may face this kind of issue.
As you said if you are not getting the prompt from console, the server might be hung.
There could be some problem with the hardware if it happens regularly.

You may issue a TC from console, other than RS. TC will generate a dump and that can be analysed later and find out the cause.

Issue a TC and call HP and get the dump analysed.

[Yogesh M Puranik]

Hi Subodh,


If server is pining,and server is unexpectedly closed network connection ,

Probability :

Please check first number of telnet seesions extend the max limit.

ie. Kernel parameter npty .

If no. of connections > npty( current value)

then u ll not be able to login to server though server is pingable.

Solution :

1]Kill some logged in sessions if you have any existing session availble anywhere else or though MP login.


Rgds

Yogesh

[subodhbagade]

Hi,

Thanks for the reply.

(1) I have checked the kernel parameter with # kmtune -q npty/nstrtel, it is giving default value 60 . also no of telnet session are less.

(2) when server was hung , i have generated the TC but it is not giving me proper booting response also dump not generated in /var/adm/crash.

(3)I have also tried the RS to reset the RST signal but during boot process it is not giving me primary boot path and giving the INIT= -7 error ( boot failure error) but after hard reboot, server detect the primary boot path. Root is not mirrored one and we don't have alternate boot path . we are facing this problem frequently and in irregular interval of time.

(4) i have executed the event monitoring cmd as
# /opt/resmon/bin/resdata -R 171966466 -r /storage/events/enclosures/ses_enclosure/0_4_0_0.8.0.255.0.15.0 -n 171966465 -a
Summary:
Enclosure at hardware path 0/4/0/0.8.0.255.0.15.0 : Hardware failure
Description of Error:
The power supply in slot B is not receiving adequate AC power.
so we have unplugged the power cable from failue path and made the server up from redundant path , but still server is in abnormal behaviour.

(5) On 16 AUG, We have applied the recommended patches ( KL,CM,NE )and upgraded the java to 1.4.2.22 successfully, and we are getting this type of issue from last 5 days, we are not sure whether we should co-relate these two thing or not ? or is it memory related issue ( not sure)

Kindly suggest and how proceed from this situation.

Thank you kindly.

Regards,
Subodh.

[Torsten.]

I'm a bit concerned about (4), the hardware problem. I don't know what hardware you have, but if a disk is just hang, the I/O is hang too and this could hang the whole server.

[subodhbagade]

Hi,

Need your valuable inputs in critical condition ;-


we are not able to detect the primary boot path, it seems to be disk fail. we don't have a root as mirror.
Also we have tried to boot from old and current kernel but no luck..

we donot having the ignite image of server ( PA-RISC 9000/800/L2000 11.23) but having the ignite image of 11.11 server( PA-RISC rp7440 ) , please confirm can we restore the os from 11.11 or it will not allow to do so.

we are getting only GSP promt , after hard rebooting and after interrupting sequence getting main prompt how we should proceed from this situation...we able to do RS but it is Continuously in bootable mode.

We are having sap application and oracle on server so is their are any way we can tried out, please suggest.


Thanks you kindly.


Regards,
Subodh.