Forum OpenACS Q&A: nssock: server ready, resuming

Collapse
Posted by Bruno Mattarollo on

Hello

I am getting these "notices" in the error log from our web server... I checked the source code and found this comment in nssock/sock.cpp, in the function SockReady. from the comment in that function it seems like a normal thing, a socket is ready to answer queries ... But should I get a Notice in the error log for that? Maybe it's something else and I am missing it ... Just curious. Here is an excerpt from the log...

[13/Sep/2002:17:10:28][12413.3076][-conn0-] Notice: nssock: server
ready - resuming
[13/Sep/2002:17:10:28][12413.6151][-conn3-] Notice: nssock: server
ready - resuming
[13/Sep/2002:17:10:28][12413.15376][-conn12-] Notice: nssock: server
ready - resuming

Collapse
Posted by Janine Ohmer on
I can't tell you what these mean, but I can tell you that when AOLserver mysteriously freezes, this is quite often the last thing in the log file.  So I think we can take from that that there is a bug or two in there somewhere.
Collapse
Posted by Bruno Mattarollo on

Thanks Janine,

Indeed we had a few days ago the first "freeze" in 83 days that the server has been up and running.

The error that I saw at that moment was the it wasn't binded to port 80 anymore but a quick killall nsd restarted the server and everything was up and running again.

I haven't had time to really look more into it but yesterday the server was quite slow (I know that this term is very subjective) and I was seeing loads of these messages in the log... So if anyone knows when this happens and what the problem might be or want to look into this, I am willing to help.

Collapse
Posted by Jonathan Ellis on
nsd/serv.c has the additional comment before running NsRunAtReadyProcs (which results in SockReady calls):
            /*
             * If this thread just free'd up the busy server,
             * run the ready procs to signal other subsystems.
             */
I don't see any correlation with crashes; these usually occur in my log after a hefty scheduled proc finally finishes running. No idea what, if anything, would cause nssock to want to wait for a scheduled proc to finish.
Collapse
Posted by Brian Fenton on
Jim Wilcockson answered this on the AOLserver mailing list:
http://www.mail-archive.com/aolserver@listserv.aol.com/msg06572.html