We've got an A400-44 server over in the Netherlands in a BT managed datacentre. unluckily for us it has no LAN console connection :-| and the server keeps deciding to just stop responding.
One minute it is fine, the next when we try to telnet to it, we just get...
Trying... Connected to nl345_4.nl.acme.com. Escape character is '^]'.
And that's it. It just hangs there. I haven't had a proper chance to investigate this yet, and until we can get someone on site to reset the server I won't be able to get on. That is unless someone has any ideas of another way which might let me get onto the box for now. SSH is not installed on the server, and it also seems that FTP is not responding...so I'm not sure if maybe inetd is dying for some reason. Just wondered if anyone had seen this kind of behaviour before and wondered what it was being caused by.
Is there any specific error messages in syslog.log ?! Try to start inetd in log mode as inetd -l to get more informations. which o/s is installed in that?
Lee, My suspects are Networks,DNS,inetd. You say "one minute it is fine..." Once you have a telnet session, can you do a local loopback? Have you reported this to the datacenter manager, it may be a local problem?
Thanks for the replies (points on their way - as soon as the "Submit Points" button starts working again instead of timing out).
OK, the box is running B.11.00 (as far as I know) and we've advised the support staff on site who are going to reboot the server.
As soon as the box is back up, I'll get on and check out syslog and a few other things.
In the meantime, allow me to hypothesise if I may...
I'm thinking that if this was inetd related it would be a "connection refused" message rather than what I'm getting.
It could be a name resolution issue, but other boxes in the same place which would be configured to use the same DNS settings are not affected.
I'm thinking this could also potentially be a problem with load...if the box is getting hammered I suppose this could be causing the problem (a sar -u should hopefully tell me this).
I don't think this is hardware related either...for example if it was the LAN card playing up, then I'd expect to lose connectivity completely, but all the way through these periods of us not being able to telnet to the server we are still able to get a response from pinging the box.
A common culprit in this situation is another machine coming on the network with the same IP address as the one you are telnetting to.
To test this, disconnect the network cable of the machine you think you are connecting to and try pinging. If you get an answer, I am right.
SEP
Steven E Protter Owner of ISN Corporation http://isnamerica.com http://hpuxconsulting.com Sponsor: http://hpux.ws Twitter: http://twitter.com/hpuxlinux Founder http://newdatacloud.com
Right, I'm back on. The box got rebooted, but didn't come up properly the first time, so the support guy on site bounced it again. Unfortunately this means that the useful syslog is no longer with us :-(
I've checked sar -u and the CPU idle % has never dropped lower than 95% since midnight last night. Current load is 0.02 0.13 0.15. No full or nearly full filesystems. Nothing in nettl to indicate a problem with duplicate IPs. Nothing of any interest in syslog.
The only thing I did find a bit odd was that the domain and nameserver arguments in /etc/resolv.conf were commented out, as was the only line in /etc/nsswitch.conf. The only line in there referenced hosts: files [NOTFOUND=continue] dns. I've uncommented these...i'm not sure why there are no entries for passwds, groups, and so on.
Hmmm...bit confused by this one, I can't really see anything that could potentially be causing the problem I'm seeing.
Did you try SEP's suggestion about the IP address being a duplicate? This random connection/not connection hang can definitely indicate that there is an IP addressing problem.
I don't think a duplicate IP is the problem here. The reason I say that, is because we never fail to get a connection as such, its just that sometimes the connection is established but then hangs and never gives you the login prompt.
Also, I ran a netfmt -f /var/adm/nettl.LOG00 and there were no entries in there to suggest a duplicate IP on the network, which is where I'd normally expect to see this logged.
