Delay in logging in with all users

[ITlearner]

Hi,
I have hpux 11.11 and the users are getting a delay in logging in.After giving uname,password they has to wait for three to five seconds for logging in(to get the prompt) but the root is able to log in immediately.Similarly the users also have delay in accessing a simple cat command also(the output takes nearly three seconds).I have checked the swapinfo ,mount points etc,and there is only a limited usage in them and they have plenty of free space.when i check syslog it is as follows.

#cat syslog.log
Aug 9 16:12:43 servername diagmond[1109]: started
Aug 9 16:12:43 servername /usr/sbin/envd[1114]: restart, and read the configuration file
Aug 9 16:13:05 servername /usr/sbin/nfsd[1396]: Setting STREAMS-HEAD high water value to 131072.



#cat OLDsyslog.log
Aug 9 16:02:59 servername snmpd[1344]: Received SNMP packet(s) from UDP: [156.151.70.88]:-12026
Aug 9 16:04:38 servername /usr/sbin/envd[1116]: terminated by signal 15
Aug 9 16:04:39 servername diagmond[1113]: Exit due to user requested abort
Aug 9 16:04:40 servername cimserver[1031]: HP WBEM Services for HP-UX stopped.
Aug 9 16:04:41 servername inetd[677]: Going down on signal 15
Aug 9 16:04:43 servername automount[1621]: exiting
Aug 9 16:04:44 servername rpcbind: terminate: rpcbind terminating on signal. Restart with "rpcbind -w"
Aug 9 16:04:44 servername syslogd: going down on signal 15

How to resolve this problem?



Regards.
Renga.V

[Chan 007]

Hi Renga,

Can you check for any disk failure either local/san or NFS - stale one.

Check dmesg any errors

ioscan for any NO_HW

Chan

[Steven E. Protter]

Shalom,

tail -f /var/adm/syslog/syslog.log

Then as a user log in.

Watch and see how fast the log gets updated and if there are messages specific to that process.

http://www.hpux.ws/system.perf.sh

You may have resource bottlenecks and those scripts will help you identify them.

The NFS idea is a good one, but not the only possible cause. Patching can be a factor as well.

SEP

[Pete Randall]

Usually this is caused by NFS mounted home directories and can be cured by adding the following to /etc/profile:

HISTFILE=/tmp/.sh_history.$(whoami)
export HISTFILE


Pete

[ITlearner]

Hi Pete,
here is the file and tell where to add it.
Regards.
Renga.V










# Insure PATH contains either /usr/bin or /sbin (if /usr/bin is not available).

if [ ! -d /usr/sbin ]
then
PATH=$PATH:/sbin

else if [ -r /etc/PATH ]
then

# Insure that $PATH includes /usr/bin . If /usr/bin is
# present in /etc/PATH then $PATH is set to the contents
# of /etc/PATH. Otherwise, add the contents of /etc/PATH
# to the end of the default $PATH definition above.

grep -q -e "^/usr/bin$" -e "^/usr/bin:" -e ":/usr/bin:"
\
-e ":/usr/bin$" /etc/PATH
if [ $? -eq 0 ]
then
PATH=`cat /etc/PATH`
else
PATH=$PATH:`cat /etc/PATH`
fi
fi
fi

export PATH

# Set MANPATH to the contents of /etc/MANPATH, if it exists.

if [ -r /etc/MANPATH ]

[ITlearner]

Hi Pete,
Heer is the /etc/profile and tell me where to add it.
Regards.
renga.v

# Insure PATH contains either /usr/bin or /sbin (if /usr/bin is not available).

if [ ! -d /usr/sbin ]
then
PATH=$PATH:/sbin

else if [ -r /etc/PATH ]
then

# Insure that $PATH includes /usr/bin . If /usr/bin is
# present in /etc/PATH then $PATH is set to the contents
# of /etc/PATH. Otherwise, add the contents of /etc/PATH
# to the end of the default $PATH definition above.

grep -q -e "^/usr/bin$" -e "^/usr/bin:" -e ":/usr/bin:"
\
-e ":/usr/bin$" /etc/PATH
if [ $? -eq 0 ]
then
PATH=`cat /etc/PATH`
else
PATH=$PATH:`cat /etc/PATH`
fi
fi
fi

export PATH

# Set MANPATH to the contents of /etc/MANPATH, if it exists.

if [ -r /etc/MANPATH ]

[V.Manoharan · Accepted solution]

Hi Renga,
you can add the HISTFILE entry in the last line. It just tells where to keep the history file.

1.Check whether the users home directoy is on nfs file system. try changing the home directory to local directory and check loging to the server.

2. Check for any issue in the nfs file system.

3.check rpcinfo -u nfs. How many wait jobs are there. also check for nfstat.

4.if it is not a production server try restarting the nfs services.

Regards
V.Manoharan

[V.Manoharan]

Hi Renga,
you can add the HISTFILE entry in the last line. It just tells where to keep the history file.

1.Check whether the users home directoy is on nfs file system. try changing the home directory to local directory and check loging to the server.

2. Check for any issue in the nfs file system.

3.check rpcinfo -u nfs. How many wait jobs are there. also check for nfsstat.

4.if it is not a production server try restarting the nfs services.

Regards
V.Manoharan

[ITlearner]

But home dirs are mounted locally and how come it gets delayed.

renga.V

[V.Manoharan]

Hi renga,
This is just to understand the issue .
1.can you tell me when the issue started .

2.Any change made in the server like patching ,hardware replacement.. changes in configuration.

3. Is the users are local users or NIS users?

4. if you do telnet 0 and try login to the same server how is the speed.

5. check for any network replated issue aswell.

you need not to worry about the messages in OLDsyslog. all are normal only.

regards
V.Manoharan

Thanks and regards
V.Manoharan

[Mel Burslan]

Just to test this, temporarily rename /etc/profile to something else like

mv /etc/profile /etc/profile.sav

and try logging in that way to see if the problem still exists.

If what you posted earlier is the entire /etc/profile, it may have been corrupted as I have seen an 'if' clause at the end which was not closed. God only knows what might have happened to /etc/profile.

Just test it to make sure.

[Michael Steele_2]

I think you've got bigger problems.

envd monitors the HW environment, ie. temperatures.

diagmond is the diagnostics daemon and also HW related.

cimserver is the common information module and translates server resource information

inetd goes down

automount fails

rpcbind then goes done.

This is a domino effect beginning with the envd daemon failing. The other daemons react to envd's failure and go down because of its absence.

Do you have SAP running on here?

I think you need to look at your patching levels as you may be a couple of years out of date. Can you verify this?

[Tim Nelson]

This certainly could be a large number of things from network, DNS, high CPU, high IO.

When you log in as root and do not have any issue is that from a physically connected console ( i.e. serial ). ?

sar 5 5 looks good, no CPU pegged ?

You will need to chop this into pieces. I have seen delayed login responses due to the login doing a reverse DNS lookup so it can put the client FQDN name in the log.

I have seen delayed login due to NFS mounts, not neccessarily the /home directory.

I have seen delayed logins due to a run away process.

Weed out all the possibilities given in these threads and I am sure you will find the solution.

[Hareesh Kumar]

Hi Renga,

You can put the lines

HISTFILE=/tmp/.sh_history.$(whoami)
export HISTFILE

just below the PATH variable in the user profile.

This will probably resolve your issue.

Regards

Hareesh
HP EMEA