Am 27.10.2009 um 01:08 schrieb Holger Amann: > > Am 27.10.2009 um 00:50 schrieb Timothee Besset: > >> I'm probably not familiar enough with how PostgreSQL works to help >> you, >> just going from my experience with Oracle here. I don't suppose >> pgsql's >> 'shared memory' could mean OS shared memory (as in, a tmpfs mount), >> because you have none. Oracle won't run if you're not giving it a >> /dev/shm tmpfs mount, but it's possible to do that for a guest (I do >> that). So all I can think of is that pgsql wants one and you haven't >> given any. >> >> TTimo > > Thank you for your help! > > I tried it (if the following is ok) and restarted Postgres, but it > made no difference in memory usage > > guest1:/var/log/postgresql# mount > /dev/hdv1 on / type ufs (defaults) > none on /proc type proc (defaults) > none on /tmp type tmpfs (size=16m,mode=1777) > none on /dev/pts type devpts (gid=5,mode=620) > tmpfs on /dev/shm type tmpfs (rw,size=4000M) > > Holger I googled around and found an information [1], that Postgres does not use POSIX SHM. "PostgreSQL is using the old SYSV shared memory and semaphores, as evidenced by ipcs[...]" So the output of ipcs on the guest is: guest1:/dev/shm# ipcs ------ Shared Memory Segments -------- key shmid owner perms bytes nattch status ------ Semaphore Arrays -------- key semid owner perms nsems ------ Message Queues -------- key msqid owner perms used-bytes messages and on a real machine: postgrestest:~# ipcs ------ Shared Memory Segments -------- key shmid owner perms bytes nattch status 0x0052e2c1 65536 postgres 600 378953728 7 ------ Semaphore Arrays -------- key semid owner perms nsems 0x0052e2c1 458752 postgres 600 17 0x0052e2c2 491521 postgres 600 17 0x0052e2c3 524290 postgres 600 17 0x0052e2c4 557059 postgres 600 17 0x0052e2c5 589828 postgres 600 17 0x0052e2c6 622597 postgres 600 17 0x0052e2c7 655366 postgres 600 17 ------ Message Queues -------- key msqid owner perms used-bytes messages So is it a Postgres specific problem, and/or could that be solved with Linux-Vserver? [1] http://www.mail-archive.com/pgsql-bugs@postgresql.org/msg17969.html