Re: vacuum verbose detail logs are unclear (show debug lines at*start* of each stage?)

Поиск
Список
Период
Сортировка
От Jeff Janes
Тема Re: vacuum verbose detail logs are unclear (show debug lines at*start* of each stage?)
Дата
Msg-id CAMkU=1xAyWnwnLGORBOD=pyv=ccEkDi=wKeyhwF=gtB7QxLBwQ@mail.gmail.com
обсуждение исходный текст
Ответ на vacuum verbose detail logs are unclear (show debug lines at *start*of each stage?)  (Justin Pryzby <pryzby@telsasoft.com>)
Ответы Re: vacuum verbose detail logs are unclear; log at *start* of eachstage; show allvisible/frozen/hintbits  (Justin Pryzby <pryzby@telsasoft.com>)
Список pgsql-hackers
On Fri, Dec 20, 2019 at 12:11 PM Justin Pryzby <pryzby@telsasoft.com> wrote:
This is a usability complaint.  If one knows enough about vacuum and/or
logging, I'm sure there's no issue.
 
| 11  DEBUG:  "t": found 999 removable, 999 nonremovable row versions in 9 out of 9 pages

I agree the mixture of pre-action and after-action reporting is rather confusing sometimes.  I'm more concerned about what the user sees in their terminal, though, rather than the server's log file.

Also, the above quoted line is confusing.  It makes it sound like it found removable items, but didn't actually remove them.  I think that that is taking grammatical parallelism too far.  How about something like:

DEBUG:  "t": removed 999 row versions, found 999 nonremovable row versions in 9 out of 9 pages  

Also, I'd appreciate a report on how many hint-bits were set, and how many pages were marked all-visible and/or frozen.  When I do a manual vacuum, it is more often for those purposes than it is for removing removable rows (which autovac generally does a good enough job of).  

Also, is not so clear that "nonremovable rows" includes both live and recently dead.  Although hopefully reading the next line will clarify that, to the person who has enough background knowledge.

 
| 12  DETAIL:  0 dead row versions cannot be removed yet, oldest xmin: 130886944
| 13  There were 0 unused item identifiers.
| 14  Skipped 0 pages due to buffer pins, 0 frozen pages.

It is a bit weird that we don't report skipped all-visible pages here.  It was implicitly reported in the "in 9 out of 9 pages" message, but I think it should be reported explicitly as well.

Cheers,

Jeff

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Tom Lane
Дата:
Сообщение: Re: TAP testing for psql's tab completion code
Следующее
От: Fabien COELHO
Дата:
Сообщение: Re: TAP testing for psql's tab completion code