On Thu, Dec 13, 2012 at 7:36 PM, Kevin Grittner <kgrittn@mail.com> wrote:
AI Rumman wrote:
I am going to restore a 6 Gb database in my development machine which is running on Centos 5.6 with memory 1 GB.
> pg_restore: out of memory > pg_restore: finished item 8570 TABLE DATA entity > pg_restore: [archiver] worker process failed: exit code 1
> I set postgresql.conf as - > shared_memory = 128 MB > maintenance_work_mem = 300 MB
> During error my OS status: > free -m > total used free shared buffers cached > Mem: 1024 975 48 0 3 857 > -/+ buffers/cache: 114 909 > Swap: 1027 0 1027 > > Please let me know what could be the actual cause of the error.
You have 1024 MB total RAM. You seem to be using 114 MB of that before starting PostgreSQL. You have PostgreSQL configured to use 128 MB of shared buffers, which is only part of its shared memory. You have configured 300 MB per maintenance_work_mem allocation. There can be several of these at one time. You are running pg_restore, which needs to use memory to interpret the map of the dump and dependencies among objects.
You are using more memory than you have.
If you really need to run PostgreSQL on a machine with 1GB of memory, you need to use a configuration much closer to the default.
Don't expect performance to be the same as on a larger server. -Kevin
