[p4] Perforce server lockup problem

Kenneth Olwing kolwing at bea.com
Mon Mar 14 02:30:30 PST 2005

> > It ended up being a memory overcommit situation that killed Linux.
> Why did you allocate so much memory? Is/was your server 
> seriously short 
> of RAM, or was there some other serious problem?

Sorry, causing confusion. It was *p4d* that was allocating the memory.
Nothing else is/was running on the machine. It was at that moment
configured with 1Gb mem + 1 Gb swap IIUC. Sure, we could & have
configured things higher, but that really just doesn't solve anything of

And, apparently it was normal for p4d to allocate all that memory in the
circumstance (haven't got the support mails handy, but as an example,
doing a sync on a client with a very loose client spec will allocate a
lot of memory).

What I'm after is mostly that a regular user process (which happened to
be p4d) could crash the OS merely by allocating memory it thinks it
needs. This is bad. Whether p4d should or should not work that way is a
different question. I'd have been happy to see it getting killed by the
OS, and hence I would have been alerted to, say, put in limitations for
MaxResults etc or do something else. Now I got a denial-of-service


More information about the perforce-user mailing list