Changes
Jump to navigation
Jump to search
###HERE###
Database Server Documentation (view source)
Revision as of 21:17, 15 March 2016
, 21:17, 15 March 2016no edit summary
checkpoint_segments=32
This is equivalent to every 512Mb
shared_buffers = 786432 100000 MB(~40% of 264 GB)
The recommendation is to use around 25-40% of RAM (we have 256Gb) for high-performance systems like ours. We might need to check that we are running 64bit and that our kernel supports this.
work_mem = 2097152500MB
#In KB, so 1024 * 512=> 524288 = 512Mb,
Note that this might be a little too high for aggresively stacked queries
But this seems to be the bite point for performance.
(kranthi: may not do much after some limit, max allowed 2GB)
maintenance_work_mem = 2621442GB (#786432 => 6 * 1073741824 => 6 * 131072 * 8 * 1024 This is how much VACUUM will use - it doesn't make a huge difference)
effective_cache_size = 1835008198GB(-3/4 of the current RAM)
#In 8k blocks. This is the maximum memory that postgres will allocate.
It is currently set to 14Gb.
###HERE###
To work as the postgres account.
