Horton

Re: Refactor pg_rewind code and make it work against a standby

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Heikki Linnakangas <hlinnaka@iki.fi>
Cc: Soumyadeep Chakraborty <soumyadeep2007@gmail.com>, Kyotaro Horiguchi <horikyota.ntt@gmail.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2020-11-15T06:48:20Z
Lists: pgsql-hackers 
Not sure if you noticed, but piculet has twice failed the
007_standby_source.pl test that was added by 9c4f5192f:

https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=piculet&dt=2020-11-15%2006%3A00%3A11
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=piculet&dt=2020-11-13%2011%3A20%3A10

and francolin failed once:

https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=francolin&dt=2020-11-12%2018%3A57%3A33

These failures look the same:

#   Failed test 'table content after rewind and insert: query result matches'
#   at t/007_standby_source.pl line 160.
#          got: 'in A
# in A, before promotion
# in A, after C was promoted
# '
#     expected: 'in A
# in A, before promotion
# in A, after C was promoted
# in A, after rewind
# '
# Looks like you failed 1 test of 3.
[11:27:01] t/007_standby_source.pl ... 
Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/3 subtests 

Now, I'm not sure what to make of that, but I can't help noticing that
piculet uses --disable-atomics while francolin uses --disable-spinlocks.
That leads the mind towards some kind of low-level synchronization
bug ...

			regards, tom lane

Commits

  1. Make pg_rewind test case more stable.

  2. Fix timing issue in pg_rewind test.

  3. Allow pg_rewind to use a standby server as the source system.

  4. pg_rewind: Refactor the abstraction to fetch from local/libpq source.

  5. pg_rewind: Replace the hybrid list+array data structure with simplehash.

  6. Refactor pg_rewind for more clear decision making.

  7. pg_rewind: Move syncTargetDirectory() to file_ops.c