How to set shmall, shmmax, shmni, etc ... in general and for postgresql

Posted by jpic on Server Fault See other posts from Server Fault or by jpic
Published on 2012-03-26T10:21:28Z Indexed on 2012/03/26 11:33 UTC
Read the original article Hit count: 721

Filed under:
|
|
|
|

I've used the documentation from PostgreSQL to set it for example this config:

>>> cat /proc/meminfo 
MemTotal:       16345480 kB
MemFree:         1770128 kB
Buffers:          382184 kB
Cached:         10432632 kB
SwapCached:            0 kB
Active:          9228324 kB
Inactive:        4621264 kB
Active(anon):    7019996 kB
Inactive(anon):   548528 kB
Active(file):    2208328 kB
Inactive(file):  4072736 kB
Unevictable:           0 kB
Mlocked:               0 kB
SwapTotal:             0 kB
SwapFree:              0 kB
Dirty:              3432 kB
Writeback:             0 kB
AnonPages:       3034588 kB
Mapped:          4243720 kB
Shmem:           4533752 kB
Slab:             481728 kB
SReclaimable:     440712 kB
SUnreclaim:        41016 kB
KernelStack:        1776 kB
PageTables:        39208 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:     8172740 kB
Committed_AS:   14935216 kB
VmallocTotal:   34359738367 kB
VmallocUsed:      399340 kB
VmallocChunk:   34359334908 kB
HardwareCorrupted:     0 kB
AnonHugePages:    456704 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:       12288 kB
DirectMap2M:    16680960 kB

>>> ipcs -l          

------ Shared Memory Limits --------
max number of segments = 4096
max seg size (kbytes) = 4316816
max total shared memory (kbytes) = 4316816
min seg size (bytes) = 1

------ Semaphore Limits --------
max number of arrays = 128
max semaphores per array = 250
max semaphores system wide = 32000
max ops per semop call = 32
semaphore max value = 32767

------ Messages Limits --------
max queues system wide = 31918
max size of message (bytes) = 8192
default max size of queue (bytes) = 16384

sysctl.conf extract:

kernel.shmall = 1079204
kernel.shmmax = 4420419584

postgresql.conf non defaults:

max_connections = 60            # (change requires restart)
shared_buffers = 4GB            # min 128kB
work_mem = 4MB              # min 64kB
wal_sync_method = open_sync     # the default is the first option
checkpoint_segments = 16        # in logfile segments, min 1, 16MB each
checkpoint_completion_target = 0.9  # checkpoint target duration, 0.0 - 1.0
effective_cache_size = 6GB

Is this appropriate ? If not (or not necessarily), in which case would it be appropriate ?

We did note nice performance improvements with this config, how would you improve it ?

How should kernel memory management parameters be set ?

Can anybody explain how to really set them from the ground up ?

© Server Fault or respective owner

Related posts about linux

Related posts about Performance