Stats not getting updated....
i am trying to identify which tables should be moved to diff drives.... i first identified the most used tables... by looking at the data in pg_statio_user_tables.....
and then i did sampling of the io-usage of these tables every 2 mins... to identify which are getting used when....
but the problem is my stats are not getting updated.... they first get updated around 20% done... and then once more at around 70% done.... why is the stats not getting updated more frequently.....
the test application i am using has a 33 min test with lots of read and write....
thx
Himanshu
---------------------------------
Discover Yahoo!
Find restaurants, movies, travel & more fun for the weekend. Check it out!
i also noted one more thing... the stats are getting updated only when i do vacuum.... is that wht is supposed to happen... or something is wrong....
thx
Himanshu
Himanshu Baweja <himanshubaweja@yahoo.com> wrote:
i am trying to identify which tables should be moved to diff drives.... i first identified the most used tables... by looking at the data in pg_statio_user_tables.....
and then i did sampling of the io-usage of these tables every 2 mins... to identify which are getting used when....
but the problem is my stats are not getting updated.... they first get updated around 20% done... and then once more at around 70% done.... why is the stats not getting updated more frequently.....
the test application i am using has a 33 min test with lots of read and write....
thx
Himanshu
---------------------------------
Discover Yahoo!
Have fun online with music videos, cool games, IM & more. Check it out!
Import Notes
Resolved by subject fallback
Himanshu Baweja wrote:
i also noted one more thing... the stats are getting updated only
when i do vacuum.... is that wht is supposed to happen... or
something is wrong....
Hmm - there's something strange going on certainly. Try a couple of
simple queries on a table and see whether the stats get updated.
It is possible that during very busy periods the stats don't get
recorded - that's part of the design.
--
Richard Huxton
Archonet Ltd
Richard Huxton <dev@archonet.com> writes:
It is possible that during very busy periods the stats don't get
recorded - that's part of the design.
IIRC, backends only ship stats to the collector at transaction commit.
Or maybe it's at the end of processing a client command. It's certainly
not continuous.
regards, tom lane
Tom Lane <tgl@sss.pgh.pa.us> wrote:
backends only ship stats to the collector at transaction commit.
Or maybe it's at the end of processing a client command. It's certainly
not continuous.
yup that i already know....
but is there any way to make it do the update more frequently.... 4 times in 30 mins... which makes the stats useless....
or there is any way to identify the usage of tables... wht i am trying to do is check the table heap_blks_read time to time.... so that i can know how much io is getting used for each table... and during which time... i am doing sampling every 2 min....
now after i have identified which tables are getting used and when... we can move them to diff partitions for better preformance.....
is there any way to know the table usage....
thx
Himanshu
---------------------------------
Do you Yahoo!?
Yahoo! Mail - You care about security. So do we.