Re: improve wals replay on secondary - Mailing list pgsql-performance

Hi Mariel,

let s keep the list in cc...

settings look ok.

what's in the recovery.conf file then?

regards,

fabio pardi

On 5/27/19 11:23 AM, Mariel Cherkassky wrote:
> Hey,
> the configuration is the same as in the primary : 
> max_wal_size = 2GB
> min_wal_size = 1GB
> wal_buffers = 16MB
> checkpoint_completion_target = 0.9
> checkpoint_timeout = 30min
>
> Regarding your question, I didnt see this message (consistent recovery
> state reached at), I guess thats why the secondary isnt avaialble yet..
>
> Maybe I'm wrong, but what I understood from the documentation- restart
> point is generated only after the secondary had a checkpoint wihch means
> only after 30 minutes or after max_wal_size is reached ?  But still, why
> wont the secondary reach a consisteny recovery state (does it requires a
> restart point to be generated ? )
>
>
> ‫בתאריך יום ב׳, 27 במאי 2019 ב-12:12 מאת ‪Fabio Pardi‬‏
> <‪f.pardi@portavita.eu <mailto:f.pardi@portavita.eu>‬‏>:‬
>
>     Hi Mariel,
>
>     if i m not wrong, on the secondary you will see the messages you
>     mentioned when a checkpoint happens.
>
>     What are checkpoint_timeout and max_wal_size on your standby?
>
>     Did you ever see this on your standby log?
>
>     "consistent recovery state reached at .."
>
>
>     Maybe you can post your whole configuration of your standby for easier
>     debug.
>
>     regards,
>
>     fabio pardi
>
>
>
>
>     On 5/27/19 10:49 AM, Mariel Cherkassky wrote:
>     > Hey,
>     > PG 9.6, I have a standalone configured. I tried to start up a
>     secondary,
>     > run standby clone (repmgr). The clone process took 3 hours and during
>     > that time wals were generated(mostly because of the
>     checkpoint_timeout).
>     > As a result of that, when I start the secondary ,I see that the
>     > secondary keeps getting the wals but I dont see any messages that
>     > indicate that the secondary tried to replay the wals. 
>     > messages that i see :
>     > receiving incremental file list
>     > 000000010000377B000000DE
>     >
>     > sent 30 bytes  received 4.11M bytes  8.22M bytes/sec
>     > total size is 4.15M  speedup is 1.01
>     > 2019-05-22 12:48:10 EEST  60942  LOG:  restored log file
>     > "000000010000377B000000DE" from archive
>     > 2019-05-22 12:48:11 EEST db63311  FATAL:  the database system is
>     starting up
>     > 2019-05-22 12:48:12 EEST db63313  FATAL:  the database system is
>     > starting up 
>     >
>     > I was hoping to see the following messages (taken from a different
>     > machine) : 
>     > 2019-05-27 01:15:37 EDT  7428  LOG:  restartpoint starting: time
>     > 2019-05-27 01:16:18 EDT  7428  LOG:  restartpoint complete: wrote 406
>     > buffers (0.2%); 1 transaction log file(s) added, 0 removed, 0
>     recycled;
>     > write=41.390 s, sync=0.001 s, total=41.582 s; sync file
>     > s=128, longest=0.000 s, average=0.000 s; distance=2005 kB,
>     estimate=2699 kB
>     > 2019-05-27 01:16:18 EDT  7428  LOG:  recovery restart point at
>     4/D096C4F8
>     >
>     > My primary settings(wals settings) : 
>     > wal_buffers = 16MB
>     > checkpoint_completion_target = 0.9
>     > checkpoint_timeout = 30min
>     >
>     > Any idea what can explain why the secondary doesnt replay the wals ?
>
>