On Tue, Aug 30, 2011 at 5:05 PM, Lonni J Friedman <netllama@xxxxxxxxx> wrote: > On Tue, Aug 30, 2011 at 2:55 PM, John R Pierce <pierce@xxxxxxxxxxxx> wrote: >> On 08/30/11 12:18 PM, Tom Lane wrote: >>>>> >>>>> total used free shared buffers cached >>>>> >> Mem: 56481 55486 995 0 15 >>>>> >> 53298 >>>>> >> -/+ buffers/cache: 2172 54309 >>>>> >> Swap: 1099 18 1081 >>>> >>>> > This is totally uninteresting. >>> >>> Yeah. You're going to need a whole lot more than 17MB of bloat before >>> it'll be possible to tell which process is at fault, given that the >>> expected process sizes are up to 10GB. >> >> indeed, its 100% normal for the linux virtual memory manager to write >> 'dirty' pages to the swapfile during idle time. > > Sure, but it shouldn't grow indefinitely? no it should not (and it can't -- you've only reserved so much space). however 35mb swap is incredibly low. on my linux workstation vm I've got 176mb used with nary an issue -- it fluctuates a lot depending what happens to be going on at the time (a workstation includes a lot of high memory low activity processes like gnome-panel which tend to swap out). had I disabled swappiness, that would have reduced memory available to do stuff with rather drastically. btw symptoms of swap thrashing are 100% full swap volume, extremely high iowait, low/no memory used for file caching, process(s) with high non shared resident memory usage, and, depending on how serious your situation is and how linux is configured, OOM errors and/or randomly killed processes. for systems not under extreme memory stress (including yours, at least at the moment) it's main purpose is to make sure some processes doesn't gobble up memory and sit on it. merlin -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
