On 2015-01-26 18:04:31 -0500, Tom Lane wrote:
> Doesn't fit this specific bug fix --- the case it addresses would cause
> already-existing rows to become unreachable from the index during crash
> recovery. AFAICS that could not create a latent problem for rows inserted
> later.
I think there were cases where it could essentially do that due to hot
pruning ending up doing strange things due to the corrupted chains. But
I don't think that's the most likely explanation.
> Still, this isn't the only bug fixed in 9.3.4/9.3.5. Personally I'm
> wondering about c0bd128c81c2b23a1cbc53305180fca51b3b61c3.
Yes, imo that bug would explain the symptoms perfectly - the cases that
lead to the discovery of that bug pretty much looked that way IIRC.
Greetings,
Andres Freund
--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services