On Tue, Oct 24, 2017 at 1:11 PM, Justin Pryzby <pryzby@telsasoft.com> wrote:
> ..which I gather just verifies that the index is corrupt, not sure if there's
> anything else to do with it? Note, we've already removed the duplicate rows.
Yes, the index itself is definitely corrupt -- this failed before the
new "heapallindexed" check even started. Though it looks like that
would have failed too, if you got that far, since the index points to
a row that does not contain the same data. (I only know this because
you dumped the heap tuple and the index tuple.)
Maybe you could try verifying a different index on the same table with
"heapallindexed", too. Perhaps that would fail in a more interesting
way.
I don't know how pg_repack works in any detail, but I have a hard time
imagining it causing corruption like this, where a single B-Tree page
is corrupt (high key invariant fails), possibly because of a torn page
(possibly due to recovery not replaying all the WAL needed, for
whatever reason).
You're using LVM snapshots -- I hope that you're aware that they're
not guaranteed to be consistent across logical volumes. There are a
few different ways that they could cause corruption like this if you
weren't careful. (In general, I wouldn't recommend using LVM snapshots
as any kind of backup solution.)
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
