Re: pg_upgrade and rsync - Mailing list pgsql-hackers

From Vladimir Borodin
Subject Re: pg_upgrade and rsync
Date
Msg-id BB18FCF0-6524-414A-863D-5CC4AF8F8B5A@simply.name
Whole thread Raw
In response to Re: pg_upgrade and rsync  (Bruce Momjian <bruce@momjian.us>)
Responses Re: pg_upgrade and rsync  (Bruce Momjian <bruce@momjian.us>)
List pgsql-hackers

2 марта 2015 г., в 21:28, Bruce Momjian <bruce@momjian.us> написал(а):

On Tue, Feb 24, 2015 at 12:13:17PM +0300, Vladimir Borodin wrote:

   20 февр. 2015 г., в 18:21, Bruce Momjian <bruce@momjian.us> написал(а):

   On Fri, Feb 20, 2015 at 09:45:08AM -0500, Bruce Momjian wrote:

           #3 bothered me as well because it was not specific enough.  I like
           what
           you've added to clarify the procedure.


       Good.  It took me a while to understand why they have to be in sync ---
       because we are using rsync in size-only-comparison mode, if they are
       not
       in sync we might update some files whose sizes changed, but not others,
       and the old slave would be broken.  The new slave is going to get all
       new files or hard links for user files, so it would be fine, but we
       should be able to fall back to the old slaves, and having them in sync
       allows that.


   Also, since there was concern about the instructions, I am thinking of
   applying the patch only to head for 9.5, and then blog about it if
   people want to test it.


Am I right that if you are using hot standby with both streaming replication
and WAL shipping you do still need to take full backup of master after using
pg_upgrade?

No, you would not need to take a full backup if you use these instructions.

Although it would be applied to documentation for 9.5 only, are these instructions applicable for upgrading from 9.3.6 to 9.4.1?

Following the instructions from patch I’ve got following errors in postgresql.log of replica after trying to start it with hot_standby = on:

< 2015-02-24 11:47:22.861 MSK >WARNING:  WAL was generated with wal_level=minimal, data may be missing
< 2015-02-24 11:47:22.861 MSK >HINT:  This happens if you temporarily set wal_level=minimal without taking a new base backup.
< 2015-02-24 11:47:22.861 MSK >FATAL:  hot standby is not possible because wal_level was not set to "hot_standby" or higher on the master server
< 2015-02-24 11:47:22.861 MSK >HINT:  Either set wal_level to "hot_standby" on the master, or turn off hot_standby here.
< 2015-02-24 11:47:22.862 MSK >LOG:  startup process (PID 28093) exited with exit code 1
< 2015-02-24 11:47:22.862 MSK >LOG:  aborting startup due to startup process failure


--
 Bruce Momjian  <bruce@momjian.us>        http://momjian.us
 EnterpriseDB                             http://enterprisedb.com

 + Everyone has their own god. +


--
Да пребудет с вами сила…

pgsql-hackers by date:

Previous
From: Shigeru Hanada
Date:
Subject: Re: Join push-down support for foreign tables
Next
From: Fabien COELHO
Date:
Subject: improve pgbench syntax error messages