pgpgout/s without swapping--what does it mean? - Mailing list pgsql-general

From Kevin Goess
Subject pgpgout/s without swapping--what does it mean?
Date
Msg-id CABZkbxiqb0_kbsA51S0hJrZFK2jt56Tt7j1Er8vMW7-WAwpebA@mail.gmail.com
Whole thread Raw
Responses Re: pgpgout/s without swapping--what does it mean?  (David Johnston <polobo@yahoo.com>)
Re: pgpgout/s without swapping--what does it mean?  (Jeff Janes <jeff.janes@gmail.com>)
Re: pgpgout/s without swapping--what does it mean?  (rahulraveendran <rahul.raveendran@au.fujitsu.com>)
List pgsql-general
We had a big increase in load, iowait, and disk i/o on a dedicated database host the other night. 

Looking at the sar logs, the problem shows itself in a big increase in pgpgout/s, which I believe is postgres paging out parts of itself to disk?

02:15:01 AM  pgpgin/s pgpgout/s   fault/s  majflt/s  pgfree/s pgscank/s pgscand/s pgsteal/s    %vmeff
02:25:01 AM     49.63    163.83    836.80      0.00    467.69      4.80      0.00      4.80    100.00
02:35:01 AM     44.85    230.29   1248.68      0.00    677.18     16.21      0.00     16.21    100.00
02:45:01 AM     47.67    190.82   1059.58      0.00    517.72      5.33      0.00      5.33    100.00
02:55:01 AM     59.00    175.58    986.87      0.00    514.08     18.13      0.00     18.13    100.00
03:05:01 AM     24.67    849.77   1382.60      0.00   1267.94     37.35      0.00     37.35    100.00
03:15:01 AM     28.67   1701.67    717.88      0.00   1231.48      0.00      0.00      0.00      0.00
03:25:02 AM     42.64  21342.02   4086.19      0.04   9701.70    415.92      0.00    414.44     99.64
03:35:01 AM     35.60  28290.69   4305.38      0.10  12906.73    623.89      0.00    615.85     98.71
03:45:01 AM     36.94  31119.30   3675.34      0.01  12456.54    527.55      0.00    521.61     98.87
03:55:01 AM     42.77  29020.72   3458.96      0.01  12165.57    557.57      0.00    553.10     99.20
Average:        41.25  11306.39   2175.59      0.02   5189.70    220.62      0.00    218.63     99.10

However, there isn't a corresponding increase in pages *in*, so if postgres is writing portions of itself out to disk, they can't be very important.  And there's no swapping going on:

02:15:01 AM  pswpin/s pswpout/s
02:25:01 AM      0.00      0.00
02:35:01 AM      0.00      0.00
02:45:01 AM      0.00      0.00
02:55:01 AM      0.00      0.00
03:05:01 AM      0.00      0.00
03:15:01 AM      0.00      0.00
03:25:02 AM      0.00      0.00
03:35:01 AM      0.00      0.00
03:45:01 AM      0.00      0.00
03:55:01 AM      0.00      0.00
   Average:         0.00      0.00

Can anybody help me understand what these statistics are suggesting, what's actually going on on this box/in postgresql?  What is it writing to disk, and why?  Is it just writing out new/changed rows, or what?

--
Kevin M. Goess
Software Engineer
Berkeley Electronic Press
kgoess@bepress.com

510-665-1200 x179
www.bepress.com

bepress: sustainable scholarly publishing  

pgsql-general by date:

Previous
From: Adrian Klaver
Date:
Subject: Re: Linux OOM-Killer
Next
From: David Johnston
Date:
Subject: Re: pgpgout/s without swapping--what does it mean?