No, there's not an easy way to see that. The /proc info, lsof, ss, etc, don't show the current listenq depth. However, there are pretty robust pieces of software that are able to determine when to spin up more theads/instances/processes without using that information. Most of them use the "I'm out of spare workers...they are all busy" condition as the trigger to add more resources..not "work is piling up at the front door" event, which is what the current listenq depth is.

Greetings, kschwab.

Let me preface the following, by letting you know, I'm not trying to be argumentative. :)

I should also note; that I'm speaking from a *BSD box. So this might explain differences.

'nuff said.
So my experiences seem to be a bit different. For example. My last log entry (somewhat abreviated):

:8090 tcpflags 0x2<SYN>; tcp_input: Connection attempt to closed port
[download]

Is fairly informative. I can also get pretty detailed information from netstat, and lsof: WAIT, LISTEN, ...

I cant speak for Linux, or OSX, for that matter (altho OSX uses BSD). But at least on BSD, it seems like I could get that sort of information. I'll look a bit deeper, and see if I can get the kind of output the OP's looking for.

--Chris

#!/usr/bin/perl -Tw
use Perl::Always or die;
my $perl_version = (5.12.5);
print $perl_version;

All interesting info. None of it is the current depth of the listenq. I suppose you could use a kernel debugger and find it, but the current number of "connections waiting for the accept() call" isn't in lsof, /proc, netstat, etc.

As it appears you're on a *NIX box. Aside from poll. The man pages for ifconfig, lsof, netstat, etc. Might give you some interesting information, that you could ask Perl to provide you in informative, and useful ways. :)

HTH

--Chris

#!/usr/bin/perl -Tw
use Perl::Always or die;
my $perl_version = (5.12.5);
print $perl_version;

This is how such an entry looks like:

1: 00000000:1A0B 00000000:0000 0A 00000000:00000002 00:00000000 00000000  1000        0 1117932 1 ffff880031e70040 100 0 0 10 0

The fist column is the local address (here a server-socket listening on port 6667) the next the remote address, then the state (here 0A for listening) then the transmit and receive queue-lengths.

Here my server has 2 connections pending.

So using the address your server listens you should be able to extract the information you need from /proc/net/tcp.

Maybe Linux::Proc::Net::TCP can help here, but I have never used it.

Awesome! Thank you. Work on Slackware too, and if it works on Slack that means it is a Linux standard. I never thought there was a file like this, perfect!