Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
Cryer, Phil C. (STL) - cont wrote: > -----Original Message----- > From: cvsnt-bounces at cvsnt.org [mailto:cvsnt-bounces at cvsnt.org] On Behalf > Of David Somers > > Cryer, Phil C. (STL) - cont wrote: >> cvs at esi1p83d:/usr/local/bin>./cvslockd -d Could not open >> /etc/cvsnt/PServer > > I guess you are using linux. > > No, Solaris 5.8 linux/solaris/whatever.... NOT windows is the important bit. > > What version of cvsnt are you using? > > cvs at esi1p83d:/export/home/cvs/pcryer>cvsnt -version > > Concurrent Versions System (CVSNT) 2.5.03 (Scorpio) Build 2382 > (client/server) > > Copyright (c) 1989-2001 Brian Berliner, david d `zoo' zuhn, > Jeff Polk, and other authors > CVSNT version (Jul 5 2006) Copyright (c) 1999-2005 Tony Hoyle and > others > see http://www.cvsnt.org > >> Initialising socket... >> Starting lock server on port 2402/tcp... >> Socket bind failed: errno 126 on socket 3 (AF 26) - closing socket >> Socket bind failed: errno 125 on socket 4 (AF 2) - closing socket >> Couldn't bind listening socket... Address already in use >> cvs at esi1p83d:/usr/local/bin> >> >> So I restarted at cvslockd -p 2403 -- is there any issue with not >> running it on the default? > > Use netstat to see who has bound to the port (i.e. netstat -ap | grep > 2402)... I've had the occasional instance of cvslockd crashing/locking > and keeping the port bound... is that's the case, pkill is your friend. > > Yep, I did that as soon as this came up, it shows a few things > listening: > >>netstat -a | grep 2402 > *.2402 *.* 0 0 24576 0 > LISTEN > *.2402 *.* 0 0 24576 0 > LISTEN > *.2402 *.* > 0 0 24576 0 LISTEN And what process has ID 24576 and, more importantly, who owns it. If nobody else is running cvs/cvsnt/cvslockd, it could be a few zombie processes... just kill them. IIRC this happened in 2.5.03... the 2.5.04 branch seems a bit better :-) FWIW, if you are getting zombie cvslockd processes and you can't upgrade to a newer version of cvsnt, set up a cron job to pkill any cvslockd processes every 24 hours or so. > The problem is, this is a shared Development box (yes, I proposed using > a new/standalone box for a CVS box, but that didn't fly). Good grief... a new box doesn't cost that much (and running your cvs system on a non-dedicated server is just asking for trouble). > This is a big > company I'm working for, and they just do not bend on many things, thus, > the box is used by a ton of people, and it's a mess. Ah, the joys of working for a one of those corps. > For now I want to > see what I need to do to tell CVSNT to look for cvslockd running on > 2403. There's an option in /etc/cvsnt/PServer to specify the port ( LockServer=localhost:2402 ... specify the port appropriate) > The other thing is they're going to do a reboot tonight (to test > someone elses' change) so perhaps whatever is running on that port won't > be there tomorrow and we can switch back, but as the startup scripts are > now set, it will come up on 2403 overnight. > > Yes. (But its probably better to find why 2402 is in use, rather than > working around the problem.) > > I agree David, but again, big company and my last day is in a few weeks, > so I'll be happy to be out of all of this 'red tape'. Hopefully I can > do everything the right way at my next position (or at least closer). Fair enough. -- David Somers typographer/programmer/whatever