Re: Sync Rep v19 - Mailing list pgsql-hackers

From Yeb Havinga
Subject Re: Sync Rep v19
Date
Msg-id 4D7007C7.4070001@gmail.com
Whole thread Raw
In response to Sync Rep v19  (Simon Riggs <simon@2ndQuadrant.com>)
Responses Re: Sync Rep v19
List pgsql-hackers
On 2011-03-03 11:53, Simon Riggs wrote:
> Latest version of Sync Rep, which includes substantial internal changes
> and simplifications from previous version. (25-30 changes).
>
> Includes all outstanding technical comments, typos and docs. I will
> continue to work on self review and test myself, though actively
> encourage others to test and report issues.
>
> Interesting changes
>
> * docs updated
>
> * names listed in synchronous_standby_names are now in priority order
>
> * synchronous_standby_names = "*" matches all standby names
>
> * pg_stat_replication now shows standby priority - this is an ordinal
> number so "1" means 1st, "2" means 2nd etc, though 0 means "not a sync
> standby".
Some initial remarks:

1) this works nice:
application_name not in synchronous_standby_names -> sync_priority = 0 (OK)
change synchronous_standby_names to default *, reload conf -> 
sync_priority = 1 (OK)

message in log file
LOG:  00000: standby "walreceiver" is now the synchronous standby with 
priority 1

2) priorities
I have to get used to mapping the integers to synchronous replication 
meaning.
0 -> asynchronous
1 -> the synchronous standby that is waited for
2 and higher -> potential syncs

Could it be hidden from the user? I liked asynchronous / synchronous / 
potential synchronous

then the log message could be
LOG:  00000: standby "walreceiver" is now the synchronous standby

3) walreceiver is the default application name - could there be problems 
when a second standby with that name connects (ofcourse the same 
question holds for two the same nondefault application_names)?

regards
Yeb Havinga




pgsql-hackers by date:

Previous
From: Andrew Sullivan
Date:
Subject: Re: Time zone database
Next
From: Tom Lane
Date:
Subject: Re: Quick Extensions Question