epmd: error in accept

[tsuraan]

> I have a system where we have multiple erlang processes running
> different programs.  It looks like somehow a race is happening where
> two epmds are started at about the same time; one of them becomes a
> happy and functional epmd, while the other one loops eternally
> printing "epmd: error in accept" to syslog, which rapidly fills up the
> file system where /var/log lives.  Looking at the epmd_srv.c file, it
> doesn't look like an error on the listensock can ever cause epmd to
> abort, which seems like a bit of a bug.  If it helps for debugging, on
> the system where I currently am seeing this, the functional epmd has
> pid 5202, while the one constants looping on its error printout has
> pid 5203.  They must have been started at almost the same time, which
> may be be a rare event in general.
>
> If I can provide any more useful information, I'd be happy to.

Looking through my messages file, I see towards the top the following
pair of messages:

Aug  2 09:38:08 localhost epmd: epmd: epmd running - daemon = 1
Aug  2 09:38:08 localhost epmd: epmd: epmd running - daemon = 1

After that, I got lines like the following until I ran out of drive space:

Aug  2 09:38:08 localhost epmd: epmd: error in accept