unable to contact rsemd at host

[Goh Mei Shyan]

Hi,

I am experiencing this problem "unable to contact rsemd at host" from time to time. Is there any method to check this? What could be the cause for this?

Thanks.

Mei Shyan

[U.SivaKumar_2]

Hi,

How many users connect to that server using remshd ?

What is the CPU load of that server ?

#top

Which process is consuming more CPU ?

regards,
U.SivaKumar

[Goh Mei Shyan]

Hi U.SivaKumar,

During the time when this error msg first occurred, the cpuload was 0.66, 1.20, 0.87; No. of user was 9.

However, when the cpuload dropped to 0.26, 0.36, 0.35 with 12 users and the highest % of cpu usage (from top): 1.18 vxfsd, 1.16 netscape, 0.97 X, I still encountered the same error.


Mei Shyan

[U.SivaKumar_2]

Hi,

Do you have only this remshd problem or any other problem related to network services ?

paste the line defining remshd from /etc/inetd.conf file of your server here.

regards,
U.SivaKumar

[Goh Mei Shyan]

Hi U.SivaKumar,

I have two lines containing remshd in the /etc/inetd.conf file.

shell stream tcp nowait root /usr/lbin/remshd remshd
kshell stream tcp nowait root /usr/lbin/remshd remshd -K

I do not have other network services problem other than this.


Mei Shyan

[Goh Mei Shyan]

Hi U.SivaKumar,

This is an add on info.

I have a cronjob running every 15mins at another server to remsh this host to check its uptime and update the uptime log. However, during the first occurence of the error message, this uptime log was not updated for the about 45mins. Could I concluded that it is due to some minimal network issue during that period of time?

The uptime log was updated after the 45mins later and it seemed fine till now.


Mei Shyan

[U.SivaKumar_2]

Hi,

check whether any instances of remsh is running ( without terminating properly)

#ps -ef | grep remsh

what is the output ?

Observe the server for another 1 hour and see whether the problem repeats.

regards,
U.SivaKumar

[Goh Mei Shyan]

Hi U.SivaKumar,

No result from ps -ef |grep remsh.

I will continue to check if I can run job without getting this error.

I was told that reboot can solve this problem. However, I hope there is a better solution for this.

Thanks,
Mei Shyan

[Jochen Heuer]

Hi,

one thing that always should be checked with inetd is if logging of inetd is enabled. If it is enabled then always a line like this is printed to your syslog for every connection:

inetd[12582]: telnet/tcp: Connection from ...

If you have this enabled make sure that nameresolution works properly. Otherwise this can lead to timeouts ...

So check your /etc/nsswitch.conf and make sure it contains an entry for hosts. If you use DNS for nameresolution too use a line like this (make sure to check /etc/hosts *first*):

hosts: files [NOTFOUND=continue] dns

Regards,

Joche

[Goh Mei Shyan]

Hi Joche,

I only found this in the /var/adm/syslog/syslog.log file :

inetd[925]: ident/tcp: Exit status 1

I can't find the line similar to
inetd[12582]: telnet/tcp: Connection from ...

and in the /etc/nsswitch.conf :

hosts: dns [NOTFOUND=continue] nis [NOTFOUND=continue] files

All workstations and servers here use dns first and they are alright.

Regards,
Mei Shyan

[Jochen Heuer]

Hello Mei,

if you don't use 'inetd -l' on the server then delays in name resolution are probably not the cause of your
login failures.

Do you use NIS? If not I would get rid of that part in /etc/nsswitch.conf. Furthermore I would still recommend to use files first instead of DNS because otherwise every lookup for 127.0.0.1 and the local ip address are sent to the nameserver ....

Regards,

Jochen