CentOS

Hi Folks,

I have a weird problem I've never run into before, and I can't seem to make heads or tails of it.

I have five identical machines, all with fresh installs of CentOS 5, with the same installation options, nothing different between them, except IP addresses (and they're sequential, 100, 101, 102, etc...).

One machine, for no apparent reason, takes a long time (10-30 seconds) when doing anything that involves authentication. If I try to log into the machine with SSH, when I enter the password, it just seems to take forever. I didn't really give it any thought, and kept going with my install. Next machine I set up went fine and normal, but the one after that exhibited the slowness problem again. It doesn't seem to be limited to SSH either, logging into any of the "slow" machines with mysql takes a long time as well. It just seems to be anything that involves authentication. These are fairly serious new machines, dual dual core, lots of memory, etc. The "normal" machines authenticate me just about as fast as I can enter the password.

A colleague of mine has run into the same problem before, but doesn't have a resolution to it, so I'm fairly certain I'm not crazy or doing something stupid (though the jury is always out on that)

I've looked at all the log files I can think of that might have some sort of clue and come up with nothing at all, comparing the "slow" machines to the "normal" ones. As far as the log files are saying, everything is AOK, things are consistently the same between all machines, and nothing glaring stands out at me.

Anybody run into this before, or have some clues as to where I could start digging?

Thanks in advance for any and all help,
-Jesse

Perhaps the machines that are slow to authenticate slow while reverse resolving the machine the login is coming from. I think I would work on that angle... make sure all the networking is set up correctly between the machines and that they can all resolve each other (and themselves) quickly.

Thanks for the suggestion, that definitely seems to be the case. I'm setting up an LVS cluster, and this only happens when a machine is configured to use the LVS router as a gateway. However, I guess I'm confused as to why this only happens on some machines configured with that gateway, and not others.

Well, back to the drawing board, thanks for the suggestion.

-Jesse

michaelnel is right on I bet. Take a couple of minutes to validate very carefully your DNS/NIS/NISPLUS/LDAP/Files that handle your intranet. It may be a simple typo. You did not mention what name resolution order you are using for hosts (look for the uncommented "host" in /etc/nsswitch.conf on all of the hosts). Normally "files" are first and that would be "/etc/hosts" most normally.