Re: lazy_truncate_heap()

Greg Stark wrote:
>
> On 31 Dec 2008, at 13:21, Simon Riggs <simon(at)2ndQuadrant(dot)com> wrote:
>>
>> Both of these bugs are minor, but the effect of either/both of them is
>> to cause more AccessExclusiveLocks than we might expect.
>>
>> For Hot Standby this means that many VACUUMs take AccessExclusiveLocks
>> on relations, which would potentially lead to having queries cancelled
>> for no reason at all.
>
> Well by default it would just cause wal to pause briefly until the
> queries with those locks finish, no?

Wait a minute. Why does an AccessExclusiveLock lead to cancelled queries
or pausing WAL application? I thought it'd just block other queries
trying to acquire a conflicting lock in the standby, just like holding
an AccessExclusiveLock on the primary does. It's unrelated to the xmin
horizon issue.

There is a noteworthy point though. In the primary, vacuum trying to
truncate takes AccessExclusiveLock conditionally, so that it doesn't
disturb queries accessing the table, and only truncates the table if it
got the lock. But in standby, we have to truncate the table, and
therefore have to acquire the lock, waiting until we get it. I guess we
have to stop applying WAL while waiting.

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com