On Fri, Jun 13, 2003 at 12:35:28PM +1000, Craig Silva wrote: > We have 2.2.7a running on redhat 9.0 with 256 MB ram. Dual cpus - but > only PII 300's. Its not running as a PDC - the PDC is an nt box. > > The system is set up as our main print server with 60 odd print queues > that service a wide variety of printers. Approx 200 odd windows pc's as > clients. The print files themselves can be largish with multi-page docs > with lots of graphics. > > What we are finding is that under certain circumstances (which appears > to be a number of big print jobs hitting the server and or users > browsing their printers) the smbd processes clog up the cpu - we were > running top and vmstat to try to get a handle on what happens when it > freezes. These processes start to take an increasing percentage of both > cpu and memory with the root owned smbd process topping out at a > reported 99% the last time the system froze. These processes hang about > in top - they don't drop out as is the case when the system is running > correctly
This is the problem that we fixed in 3.0 by re-designing the printer job query code to be (mostly) asynchronous. The branch APPLIANCE_HEAD contains these fixes in a 2.2.x codebase. I'd recommend trying out the beta version of 3.0 to see if this fixes the problem. Jeremy -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
