possible problem with indexes postgresql-6.5.2

Поиск
Список
Период
Сортировка
От Jim Mercer
Тема possible problem with indexes postgresql-6.5.2
Дата
Msg-id 19991108142432.T15182@reptiles.org
обсуждение исходный текст
Список pgsql-general
i have a large table that gets added to frequently (2-3 times per hour, 50K
or more records each time).

this table has a single index, unique, a char(20) field.

every now and then, the index seems to grow abnormally, although it appears
to function fine.

here is an ls -l, before:

-rw-------  1 pgsql  wheel  1073741824 Nov  6 23:34 gt
-rw-------  1 pgsql  wheel  1073741824 Nov  6 23:36 gt.1
-rw-------  1 pgsql  wheel  1073741824 Nov  6 23:37 gt.2
-rw-------  1 pgsql  wheel   796180480 Nov  7 17:11 gt.3
-rw-------  1 pgsql  wheel           0 Oct 28 00:15 gt.4
-rw-------  1 pgsql  wheel  1073111040 Nov  7 17:11 gt_callid_key

another round of inserts hit, and then i did:
"drop index gt_callid_key;"
"vacuum verbose;"
"create unique index gt_callid_key on gt (callid);"

-rw-------  1 pgsql  wheel  1073741824 Nov  7 17:22 gt
-rw-------  1 pgsql  wheel  1073741824 Nov  7 17:24 gt.1
-rw-------  1 pgsql  wheel  1073741824 Nov  6 23:37 gt.2
-rw-------  1 pgsql  wheel   797941760 Nov  7 18:12 gt.3
-rw-------  1 pgsql  wheel           0 Oct 28 00:15 gt.4
-rw-------  1 pgsql  wheel   359399424 Nov  7 18:12 gt_callid_key

from time to time, i have seen gt_callid_key grow to in excess of 50% the size
of the main table.

any ideas?

this is postgres-6.5.2 (patched from 6.5.1) on intel hardware under FreeBSD.

--
[ Jim Mercer                 jim@reptiles.org              +1 416 506-0654 ]
[          Reptilian Research -- Longer Life through Colder Blood          ]
[  Don't be fooled by cheap Finnish imitations; BSD is the One True Code.  ]

В списке pgsql-general по дате отправления:

Предыдущее
От: Martin Weinberg
Дата:
Сообщение: Memory leak in libpq & libpq++?
Следующее
От: The Hermit Hacker
Дата:
Сообщение: Re: [GENERAL] Banner (fwd)