[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Delay in connecting to service



Marcin Owsiany <porridge@pandora.info.bielsko.pl> writes:
> On Fri, Oct 06, 2000 at 09:56:44AM +0200, Fredrik Liljegren wrote:
> > Oct  2 08:52:04 bifrost sshd[228]: debug: Forked child 18444.
> > Oct  2 08:53:19 bifrost sshd[18444]: Connection from 10.10.11.5 port 2038

> sshd is waiting for DNS lookup timeout

I don't think so, because that is another delay that we've had earlier
but dont have anymore, like the following:

Aug 24 17:16:54 bifrost sshd[478]: Accepted rsa for tombe from 10.10.12.5 port 2310
Aug 24 17:17:09 bifrost sshd[478]: Could not reverse map address 10.10.12.5.

This comes after the authentication has started, not as the above,
between the fork and the connection to the new child.

Is there any other DNS-lookup that is done before the child is
connected to the port, that can explain the delay I get when I start
it from inetd also?


> BTW: This is probably not best place to ask such questions. Use
> debian-isp instead...

Because they would know more?  I cc them and include the original
message..  The service is internal in our company instranet and has
nothing else to do with isp's.

Regards, Fredrik


--- Begin Message ---
I have two Debian Potato servers, that I among other things connect to
via ssh.
  In both these, there is SOMETIMES a dalay of over a minute between
the connection is established/noticed, and the actual
authentication-process begins.  For example:

Oct  2 08:52:04 bifrost sshd[228]: debug: Forked child 18444.
Oct  2 08:53:19 bifrost sshd[18444]: Connection from 10.10.11.5 port 2038

The time between the fork (when the connection is established
according to my ssh-client) and the childs "Connection" is mostly less
than a second, but sometimes, as in this case, over a minute.

I guess this is some setup-problem.  I know other services get this as
well, as it does even when I start sshd from inetd.  These servers are
not heavily loaded; mostly there's only one connection at a time.  The
strange thing is that this occures even on a newly-installed potato
server with everything configured "out of the box" (allthough I know
of potato-systems that do not have this problem).

I have tried recompiling the kernel, with no effect (I now use 2.2.17
on one box and Debians image of 2.2.14 on the other).  All the
programs are potatos last version as of today (sshd, glibc, inetd
etc).  The networking is done with eepro100.  The main server is a PII
400MHz with 384 MB memory and 498004k swap-space.  Both use samba
exports.  I have checked kern.log and the other logs without finding
any problem-reports that could have anything with this to do.  I have
run memory-checks with memtest86 (thoroughly from boot).  APM is
disabled.  The load-avarage is normally below 0.40.  The hard disk is
a scsi divided into 2 GB for root and 7GB-partitions for our
cvs-archives and home.  One of the servers use pam-limits, the other
doesn't.  Both systems use pam.  They have static ip.

..and I have absolutely no idea left what I could do other than
switching to another distribution (which I don't want...).

With hope for help, Fredrik


-- 
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org



--- End Message ---

Reply to: