Dear List,
Today has been good since morning. Although it is a lean day
for us but the indications are nice. I thank everyone who shared
the concern. I think the most significant change has been to reduce
shared_buffers from 10G to 4G , this has lead to reduced memory
usage and some breathing
Dear List ,
A simple (perl) script was made to 'watch' the state transitions of
back ends. On startup It captures a set of pids for watching
and displays a visual representation of the states for next 30 intervals
of 1 seconds each. The X axis is interval cnt, Y axis is pid and the
origin is on
A scary phenomenon is being exhibited by the server , which is the server
is slurping all the swap suddenly , some of the relevant sar -r output are:
10:30:01 AM kbmemfree kbmemused %memused kbbuffers kbcached
kbswpfree kbswpused %swpused kbswpcad
10:40:01 AM979068 31892208 97.02
On Fri, 2010-06-25 at 15:25 +0530, Rajesh Kumar Mallah wrote:
shared_buffers = 10GB # min 128kB
work_mem = 512MB# min 64kB
These are still pretty high IMHO. How many *concurrent* connections do
you have?
--
Devrim GÜNDÜZ
PostgreSQL
Rajesh Kumar Mallah wrote:
A scary phenomenon is being exhibited by the server , which is the server
is slurping all the swap suddenly
8 1 4192912 906164 6100 2787364000 2277 858 13440 16235
63 8 19 10 0
I reduced work_mem from 4GB to 2GB to 512MB (now). I clearly remember
Dear List,
Hmmm , lemme test efficacy of pg_tune.
I would reduce shared buffers also.
regarding concurrent queries:
its now non business hours and
SELECT procpid,current_query from pg_stat_activity where
current_query not ilike '%idle%' ;
is just 5-10, i am yet to measure it during business
Rajesh Kumar Mallah mallah.raj...@gmail.com wrote:
its now non business hours and
SELECT procpid,current_query from pg_stat_activity where
current_query not ilike '%idle%' ;
is just 5-10, i am yet to measure it during business hours.
Be careful about 'IDLE in transaction' status. Those
I changed shared_buffers from 10G to 4G ,
swap usage has almost become nil.
# free
total used free sharedbuffers cached
Mem: 32871276 245758248295452 0 11064 22167324
-/+ buffers/cache:2397436 30473840
Swap: 4192912
Dear List,
pgtune suggests the following:
(current value are in braces via reason) , (*) indicates significant
difference from current value.
default_statistics_target = 50 # pgtune wizard 2010-06-25 (current 100
via default)
(*) maintenance_work_mem = 1GB # pgtune wizard 2010-06-25 (16MB
On 25/06/10 16:59, Rajesh Kumar Mallah wrote:
when i reduce max_connections i start getting errors, i will see again
concurrent connections
during business hours. lot of our connections are in IDLE in
transaction state during business
this peculiar behavior of mod_perl servers have been
Rajesh Kumar Mallah mallah.raj...@gmail.com wrote:
pgtune suggests the following:
(current value are in braces via reason) , (*) indicates
significant difference from current value.
Different people have come to different conclusions on some of these
settings. I believe that's probably
Rajesh Kumar Mallah wrote:
default_statistics_target = 50 # pgtune wizard 2010-06-25
(current 100 via default)
(*) effective_cache_size = 22GB # pgtune wizard 2010-06-25 (18GB ,
specified)
checkpoint_segments = 16 # pgtune wizard 2010-06-25 (30 , specified)
You probably want to
Dear Greg/Kevin/List ,
Many thanks for the comments regarding the params, I am however able to
change an
experiment on production in a certain time window , when that arrives i
shall post
my observations.
Rajesh Kumar Mallah.
Tradeindia.com - India's Largest B2B eMarketPlace.
13 matches
Mail list logo
