telnetd getpid: peer died : bad file number (yet again!)

[Steve Howie]

Environment:

HP-UX 10.20, K200 server, 256Mb RAM

get periodic errors of this form:

telnetd[xxx] getpid: peer died : bad file number

Simultaneous to this, a telnet connection cannot be initiated and times out. Have already applied PHNE 13414 (telnetd cum. patch), and increased npty and nstrpty to 800 - all to no avail.

Any ideas?

Thanks,

Steve Howie showie@uoguelph.ca
University of Guelph

[Patrick Wallek]

Have you checked your maxfiles, nfile and nproc kernel parameters to see if maybe you are running up against the limit on those?

I think it's a stretch, but it's worth a shot.

[Steve Howie]

Thanks for replying ..

I've checked these parms (except maxfiles) using Glance and they're very overallocated right now ... 18-20% utililization at most ... so this is probably not the problem

soft limit on open files is 60, hard is 1024 ( from SAM)) maxfiles and maxfiles_lim ...

Scotty

[Steven Sim Kok Leong]

Hi,

Just a hunch. Have you also checked maxuprc?

Hope this helps. Regards.

Steven Sim
Brainbench MVP for Unix Admin
http://www.brainbench.com

[Rita C Workman]

Generally you get telnet getpid: peer died errors because.
1. Missing ARPA patches (but you've checked that)
2. Too low on NTPY etc parms (but you up'd them)
3. You do not have an internal DNS server, so your telnet sessions go past the hosts file and then to your external DNS, which then takes about 100 seconds to respond with a login - Now if your tmout is set too small you never get the login...But I think your ok there too..

...But in your case it says something about a Bad File Number. And this is something that normally doesn't come with this type of error.
SOOOOOOO....I'm thinking maybe your seeing these because there is a problem in your kernel. I would recommend creating a new system file the existing kernel with a:
a) usr/lbin/sysadm/system_prep -s system
b) and then make sure your ntpy and nstrpy are set to 800 in your system file manually
c) continue to build the new kernel from this point and reboot.

Just a thought,