Chris, Is this linux? Give output of /proc/meminfo I noticed this when we had allocated hugepages (rhel4 u8) and because of incorrect ulimit setting it was not being used. As OS has already reserved the memory for hugepages , database took rest memory and also consumed lot of swap. I believe you too are facing same issue Regards Amit http://askdba.org/weblog/ On Tue, Jul 27, 2010 at 10:44 PM, David Roberts < big.dave.roberts@xxxxxxxxxxxxxx> wrote: > If you are questioning why the swap space is filling up before memory has > been exhausted; then there are various Unix variants that write the > executable image to swap at the same time that the program is run. > > My understanding is that this should result in a greater contiguity with > regards the running code and the image on disk and should lead to > performance improvements and the code is swapped out to disk. > > I believe that it's an ELF thing. although there is no theoretical reason > why COFF based systems shouldn't implement the same functionality. > > > Regards > > Dave > > On Tue, Jul 27, 2010 at 5:58 PM, Stephens, Chris > <Chris.Stephens@xxxxxxx>wrote: > >> I got a call from the system administrators yesterday about excessive >> swap usage on one of our database servers. Swapping was not occurring but a >> significant percentage of swap was allocated. >> >> >> >> I recently created an 11.2.0.1 database on the server and configured the >> new memory_target and memory_max_target parameters to 2G. The odd thing is >> that free –m showed a difference of 8GB in memory allocation between the >> database being up and down. >> >> >> >> Reading up on the new auto-memory magic a little indicates that the new >> memory management makes use of /dev/shm. >> >> >> >> Oddly (to me anyways) that is configured to 8gb: >> >> >> >> $ df -k /dev/shm >> Filesystem 1K-blocks Used Available Use% Mounted on >> tmpfs 8153816 625852 7527964 8% /dev/shm >> >> >> >> >> >> I’m struggling with how to explain this. >> >> >> >> That also got me thinking about the usefulness of an a script that >> accurately maps the memory usage of each oracle instance on a particular >> server. In my short reading, it sounds like ‘ps’ isn’t very reliable since >> the memory reported per process also includes shared memory? So, any >> thoughts on the most accurate way to break down the memory consumption of an >> instance and all user processes connected to it? >> >> >> >> Thanks in advance for any insight. >> >> CONFIDENTIALITY NOTICE: >> This message is intended for the use of the individual or entity to which >> it is addressed and may contain information that is privileged, confidential >> and exempt from disclosure under applicable law. If the reader of this >> message is not the intended recipient or the employee or agent responsible >> for delivering this message to the intended recipient, you are hereby >> notified that any dissemination, distribution or copying of this >> communication is strictly prohibited. If you have received this >> communication in error, please notify us immediately by email reply. >> >> >> >