Horton

Re: error context for vacuum to include block number

Justin Pryzby <pryzby@telsasoft.com>

View thread View on postgr.es
From: Justin Pryzby <pryzby@telsasoft.com>
To: Masahiko Sawada <masahiko.sawada@2ndquadrant.com>
Cc: Alvaro Herrera <alvherre@2ndquadrant.com>, Andres Freund <andres@anarazel.de>, Michael Paquier <michael@paquier.xyz>, pgsql-hackers@postgresql.org
Date: 2020-03-04T21:51:59Z
Lists: pgsql-hackers

Attachments

On Wed, Mar 04, 2020 at 04:21:06PM +0900, Masahiko Sawada wrote:
> Thank you for updating the patch. But we have two more places where we
> do fsm vacuum.

Oops, thanks.

I realized that vacuum_page is called not only from lazy_vacuum_heap, but also
directly from lazy_scan_heap, which failed to update errcbarg.  So I changed to
update errcbarg in vacuum_page.

What about these other calls ?  I think granularity of individual function
calls requires a debugger, but is it fine issue if errors here are attributed
to (say) "scanning heap" ?

GetRecordedFreeSpace
heap_*_freeze_tuple
heap_page_prune
HeapTupleSatisfiesVacuum
LockBufferForCleanup
MarkBufferDirty
Page*AllVisible
PageGetHeapFreeSpace
RecordPageWithFreeSpace
visibilitymap_*
VM_ALL_FROZEN

> These functions have LVDeadTuples and LVRelStats but LVDeadTuples can
> be referred by LVRelStats. If we want to use LVRelStats as callback
> argument, we can remove function arguments that can be referred by
> LVRelStats.

That doesn't work easily with parallel vacuum, which passes not
vacrelstats->dead_tuples, but a dead_tuples pulled out of shm_toc.

But it was easy enough to remove "reltuples".

-- 
Justin

Commits

  1. Avoid calls to RelationGetRelationName() and RelationGetNamespace() in

  2. Introduce vacuum errcontext to display additional information.

  3. Fix mesurement of elapsed time during truncating heap in VACUUM.

  4. Allow vacuum command to process indexes in parallel.

  5. Refactor code dedicated to index vacuuming in vacuumlazy.c

  6. Remove duplicated progress reporting during heap scan of VACUUM