Patrick Goldbronn - SFME/LGLS wrote:
> 
> > Each time you run a cvs command, it opens a connection to the server, and the
> > number of times you're calling cvs is over the server's connection limit. Some
> > solutions/workarounds:
> 
> Right, but this connections are not in parallel, so there is one
> connection at one time (I suppose ! Perhaps shell run in parallel ?).

AFAIK, you can't assume that the connection is closed when the process exits.
There is a delay as the connection state progresses from connected to closed.
Parallel connections would only increase the number of concurrent connections,
and you'd see the error sooner.

To see what I mean, run netstat and your script at the same time to monitor the
connection states.

> > 1. Avoid putting cvs calls in loops. Instead, do something like 'cvs log
> > $files_list' or 'find files |xargs cvs log'
> 
> no because I want information for one file and I make something if it
> has some information

If you want info for one file, why is the log call in a loop? And what's
stopping you from parsing more than one log entry at a time? If you must call
cvs each time, you'll have to include a delay or do something else to keep the
connection count below the limit.

> I have put in my inetd.conf "nowait.120" and all work fine !!!!

I avoided listing this option in my previous post because it will only delay the
problem. When you have a sequence of calls large enough to hit the new limit,
you'll be stuck again.

-Matt

_______________________________________________
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs

Reply via email to