]> granicus.if.org Git - postgresql/commit
Avoid unnecessary failure in SELECT concurrent with ALTER NO INHERIT.
authorTom Lane <tgl@sss.pgh.pa.us>
Fri, 12 Jan 2018 20:46:37 +0000 (15:46 -0500)
committerTom Lane <tgl@sss.pgh.pa.us>
Fri, 12 Jan 2018 20:46:37 +0000 (15:46 -0500)
commitc2a7044a5128a17256c771b545a534c7dccb7736
tree88bfe1f524b646471bcd1ef2adaa8c0d53a66eba
parent6520d4a9692882a7d5233dac40c4b7ff07f55049
Avoid unnecessary failure in SELECT concurrent with ALTER NO INHERIT.

If a query against an inheritance tree runs concurrently with an ALTER
TABLE that's disinheriting one of the tree members, it's possible to get
a "could not find inherited attribute" error because after obtaining lock
on the removed member, make_inh_translation_list sees that its columns
have attinhcount=0 and decides they aren't the columns it's looking for.

An ideal fix, perhaps, would avoid including such a just-removed member
table in the query at all; but there seems no way to accomplish that
without adding expensive catalog rechecks or creating a likelihood of
deadlocks.  Instead, let's just drop the check on attinhcount.  In this
way, a query that's included a just-disinherited child will still
succeed, which is not a completely unreasonable behavior.

This problem has existed for a long time, so back-patch to all supported
branches.  Also add an isolation test verifying related behaviors.

Patch by me; the new isolation test is based on Kyotaro Horiguchi's work.

Discussion: https://postgr.es/m/20170626.174612.23936762.horiguchi.kyotaro@lab.ntt.co.jp
src/backend/optimizer/prep/prepunion.c
src/test/isolation/expected/alter-table-4.out [new file with mode: 0644]
src/test/isolation/isolation_schedule
src/test/isolation/specs/alter-table-4.spec [new file with mode: 0644]