Redhat Enterprise 4 update 4 is out . -----Original Message----- From: redhat-list-bounces@xxxxxxxxxx [mailto:redhat-list-bounces@xxxxxxxxxx] On Behalf Of Carter, Richard (OMC) Sent: Wednesday, September 06, 2006 3:53 AM To: redhat-list@xxxxxxxxxx Subject: AS2.1 - low memory tuning - enomem errors (code 12) Hi, I'm running AS2.1 and I'm getting 'enomem - code -12' returned to some applications, including Oracle ocfs. At the time I have lots of 'free' memory, as shown with the free command. I've read some forum notes which suggest that it is an issue with the allocation of 'low memory', does this seem right? If so, has anyone got any thoughts on: - how to monitor the low memory usage? - best parameters to tune to try and lessen the enomem errors? things I've seen suggested to monitor: - /proc/meminfo - - The low memory size number seems to be set at around 800M. (can this be changed?) - The low mem free seems to vary from 200M to 2M. - Where it the first 600M go? - Is the usage from 200M to 2M reflected in 'slab' usage, see below - /proc/slabinfo - This seems to show you the 'slabs', which I think are caches set up in low memory. - Does lowmem total - total of slab sizes = lowmem free? (after an initial piece has been allocated? 500M?) parameters to tune: I've seen suggested: - max_map_count (currently 65535) - seems to be the number of memory regions a process can use? - is there a way to show how many regions a process has mapped? - try lowering? - bigpages=Xmb on the grub line. - /proc/meminfo shows 'bigpage usage'. Do these help, particularly with oracle ocfs? Or Oracle in general? - pagecache - are these pages kept in 'low memory'? - freepages - does this effect low memory usage? Thanks for any help. -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=subscribe https://www.redhat.com/mailman/listinfo/redhat-list -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
