Question about process scheduling - revisited.

[Lynch, Harold]

I originally had a question about the way cvs server process get scheduled.

Let me elaborate.

The setup:

        Server:
        DELL dual intel processor (2.0 gig hertz).
        4 gig ram
        500 gig harddriver.
        Dual gigabit and 10t network interfaces.

        CVS 1.11.15 server on this machine (pserver is the connection method).

Originally I thought it was a checkout time issue. But now I can see the occasional "slow rtag".

Normally it takes between 4 and 15 mintes to pull a "sandbox", or about 15 minutes to "rtag" the same
module.

Occasionally we are seeing the checkout time jump to over 1 hour, or the rtag to take 2 hours.

When I look at the server process on that corresponds to this "slow moving" command, I see that it
seems to intermittenly get high then low cpu usage, and the total cpu time for that process keeps
getting higher. But I've never seen it (in top) total peg the cpu's.

Now this repository is about 3.5 years old, and quite large (over 40,000) files, lots of branches, etc.

We've had the network people monitor that end, and they aren't seeing anything lost there.

Does anybody have any ideas on what to look for here.
Or is there pertanant information I have left out?

Harold Lynch