>>> On 5/23/2013 at 06:26 PM, Helio Mario Neves Pimentel de Oliveira <[email protected]> wrote: > Hi listers, > > Does anybody have an idea/suggestion/solution to what I describe below? > > IBM Z10 LPAR running SUSE SLES 10 SP4 > > Other day some external applications couldn*t connect to the data base, > neither to the SMB, neither do SSH (console) running on it. > All appls stayed hanging for about 20 minutes. > All processing was resumed without further notice AND WITHOUT ANY CORRECTIVE > ACTION.
I've seen cases where the size of the SGA and PGA, combined with the number of process mmapping it and remote network connections to the database, just about exhausted all the available "physical" memory defined for the system. In those cases, the kernel went into a mode where it was trying to find pages to page out, but almost everything was marked non-pageable, so it couldn't. After a while, connections would time out, processes would die, freeing up some memory, and the kernel would start to be able to do some paging and things would recover. Because of the nature of the circumstances, no advance warning was possible because no ramp up in paging was happening prior to the critical point where everything just ground to a halt. You might want to go through the exercise of adding up all the things using memory that will cause it to not be paged out and see if you're too close to the amount of memory available to the system. Mark Post ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/
