topic Re: who -r not working in Operating System - HP-UX

who -r not working

Roman Schmidt — Fri, 20 Mar 2009 21:44:56 GMT

hello,

i have a hp-ux B.11.11 machine here where the command who -r gives an empty output, return value is zero.

the who command itself does work, the /etc/inittab and /sbin/init seem OK.

where could i start to search for the problem?

is there another way to see the current runlevel?

Re: who -r not working

Mark McDonald_2 — Fri, 20 Mar 2009 22:40:42 GMT

you may be able to work out the current run lever from the last entries in the /etc/rc.log and compare these with what is in /etc/rc?.d

Can you post the output of:
whence who

Re: who -r not working

Roman Schmidt — Fri, 20 Mar 2009 22:43:13 GMT

'whence who' gives me:

/usr/bin/who

Re: who -r not working

Roman Schmidt — Fri, 20 Mar 2009 22:44:43 GMT

btw.

the last 100 entries '/etc/rc.log' are:

AxInitLocale: Error occurred during locale and conversion initialization.

Re: who -r not working

Dennis Handly — Fri, 20 Mar 2009 22:55:03 GMT

If "who -r" doesn't work, /etc/utmp has been corrupted/emptied.

See these other threads about runlevel:
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1239710
INIT_STATE=s

http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1238202
who -r not working in single user mode.

It also mentions the command getrunlvl.

Re: who -r not working

Roman Schmidt — Fri, 20 Mar 2009 23:17:40 GMT

hello, thanks for your reply, but still i don't get the point.

(note: i won't be able to reboot / change runlevels of the machine because it's a customer's server)

getrunlvl doesn't give me any output either, the machine wasn't rebooted for 200 days now and /etc/utmp isn't empty.

'strings /etc/utmp' gives me:

root@asbd4z :/root/home/root :strings /etc/utmp
root
consconsole
ts/1pts/1
ts/2pts/2
ts/3pts/3

B<ts/4pts/4
pB<ts/5pts/5

B<ts/6pts/6

ts/7pts/7
ermclnt
ts/0pts/0

Bpts/ta

pts/1
I8S5
LOGIN
pts/2

i replaced every ip with and every username with . except that it's exactly the content of the file.

maybe it has something to do with that storm of:

AxInitLocale: Error occurred during locale and conversion initialization.

in the /etc/rc.log? (although that file was last modified over 3 months ago)

Re: who -r not working

Dennis Handly — Fri, 20 Mar 2009 23:41:17 GMT

>I won't be able to reboot / change runlevels of the machine because it's a customer's server

Since the machine is up, your runlevel is fine. Is your problem some script that insists on checking the runlevel and that's failing?

>getrunlvl doesn't give me any output either

Hmm, I forgot what that looks at. You could use tusc to check.

>the machine wasn't rebooted for 200 days now and /etc/utmp isn't empty.

It could be corrupted.
What does this show:
/usr/sbin/acct/fwtmp < /etc/utmp

Re: who -r not working

Roman Schmidt — Sat, 21 Mar 2009 00:09:23 GMT

seems that the customer needs that command, he asked us to fix this issue.

here the output of '/usr/sbin/acct/fwtmp < /etc/utmp':

root cons console 8064 7 0000 0001 1219995548 Aug 29 09:39:08 2008
ts/1 pts/1 2278 8 0000 0000 1229333766 Dec 15 10:36:06 2008
ts/2 pts/2 10408 8 0000 0000 1235750764 Feb 27 17:06:04 2009
ts/3 pts/3 29760 7 0000 0000 1237570989 Mar 20 18:43:09 2009
ts/4 pts/4 2157 7 0000 0000 1237571440 Mar 20 18:50:40 2009
ts/5 pts/5 2986 7 0000 0000 1237592505 Mar 21 00:41:45 2009
ts/6 pts/6 7449 8 0000 0000 1237309059 Mar 17 17:57:39 2009
ts/7 pts/7 27854 8 0000 0000 1229612171 Dec 18 15:56:11 2008
ermclnt 22780 8 0000 0000 1237569841 Mar 20 18:24:01 2009
ts/0 pts/0 14496 7 0000 0000 1237200158 Mar 16 11:42:38 2009
LOGIN ta pts/ta 1910 8 0000 0000 1233141984 Jan 28 12:26:24 2009
1 pts/1 22654 8 0000 0000 1228428085 Dec 4 23:01:25 2008
LOGIN 2 pts/2 25517 8 0000 0000 1233140892 Jan 28 12:08:12 2009

i replaced the usernames by und the ip's with .

i've attached the output of 'tusc who -r' as well. (i've X'd out the usernames).

Thanks for the help.

Re: who -r not working

Patrick Wallek — Sat, 21 Mar 2009 01:01:09 GMT

It looks like pwgr (password & group caching daemon) is running. Try stopping this and then running your 'who -r' again.

# /sbin/init.d/pwgr stop

# who -r

If you want to restart it:

# /sbin/init.d/pwgr start

You could then try the 'who -r' again and see if just stopping and restarting pwgr "fixed" the problem.

No, stopping pwgr will not prevent anyone from logging in and should not have any other adverse effects.

I never run this on any of my servers. From what I have read it is really only useful if there are a lot of users logging in and out.

Re: who -r not working

Roman Schmidt — Sat, 21 Mar 2009 01:06:41 GMT

yes, this service is running.

root@xxxxxx :/root/home/root :/sbin/init.d/pwgr stop
pwgrd stopped
root@xxxxxx :/root/home/root :who -r
root@xxxxxx :/root/home/root :/sbin/init.d/pwgr start
pwgrd started
root@xxxxxx :/root/home/root :who -r
root@xxxxxx :/root/home/root :

no change :(

Re: who -r not working

Roman Schmidt — Sat, 21 Mar 2009 01:18:51 GMT

well, i think i've got some idea what could it be. if i study the tusc output and compare that to the one of another hpux machine i seems that 'who -r' searches for following entries in /etc/utmp:

system boot
run-level

so if the entries would be like:

system boot
run-level 3

then getrunlvl shows:
New_level 3 Old_level S

i can't find both of these entries in /etc/utmp on the machine with issues.
it really seems that this file got messed up (or deleted some time ago).

is there a way to make this file valid again without a reboot?

Re: who -r not working

Patrick Wallek — Sat, 21 Mar 2009 01:35:40 GMT

>>is there a way to make this file valid again
>>without a reboot?

Not that I know of.

Re: who -r not working

Roman Schmidt — Sat, 21 Mar 2009 01:40:09 GMT

ok thanks for help. i guess we have to live with that until the next reboot.

Re: who -r not working

Dennis Handly — Sat, 21 Mar 2009 04:08:17 GMT

>is there a way to make this file valid again without a reboot?

Sure, write a program.
Since you are on 11.11 you probably don't need to do that.

>Patrick: Not that I know of.

Right, there is no specific command(s) that will fix it. And even harder on 11.23 since there is utmps & utmpd.

>I guess we have to live with that until the next reboot.

(See easier way below, first.)
If you are ambitious you might try the following:
1) Back up /etc/utmp in case we make it worse.
2) You might be able to extract the info from a good system with dd(1).
3) Then use "cat >> /etc/utmp" to append to the end.
4) They try "who -r".

>if I study the tusc output and compare that to the one of another hp-ux machine it seems that 'who -r' searches for following entries in /etc/utmp:
system boot run-level

Of course. See utmp(4):
http://docs.hp.com/en/B2355-60130/utmp.4.html

Just use tusc on a good system to compute where the good record is, then use dd to extract it.

In your tusc dump on the bad system, suppose this is the entry you want to copy:
read(5, "x x x x x x x \0t s / 7 p t s / ".., 60)
Counting from 0, this is record 7:
dd if=/etc/utmp of=runlevel bs=60 iseek=7 count=1

Then just copy to the bad system and do:
cat runlevel >> /etc/utmp

Actually since it is /etc/utmp you can do:
1) On good system:
2) /usr/sbin/acct/fwtmp < /etc/utmp | grep "run-level " > runlevel
3) Copy runlevel to bad system
4) On bad system
5) /usr/sbin/acct/fwtmp < /etc/utmp > utmp.txt
6) cat runlevel >> utmp.txt
7) cp /etc/umtp utmp.save
8) /usr/sbin/acct/fwtmp -ic < utmp.txt > /etc/utmp

>I can't find both of these entries in /etc/utmp on the machine with issues.

Yes, someone emptied out /etc/utmp and lost that entry.

Re: who -r not working

Roman Schmidt — Sat, 21 Mar 2009 04:21:40 GMT

hi,

i'll see what the customer says on monday and if he insist i'll try to recover the file.

thanks for the info :)

Re: who -r not working

Roman Schmidt — Tue, 26 Apr 2011 02:46:36 GMT

closing