ERROR: found xmin from before relfrozenxid

Поиск
Список
Период
Сортировка
От Mariel Cherkassky
Тема ERROR: found xmin from before relfrozenxid
Дата
Msg-id CA+t6e1kBb0-h_B_hZpab0uOgiE2Go_VLnayd9RkHaX26ybdUmw@mail.gmail.com
обсуждение исходный текст
Ответы Re: ERROR: found xmin from before relfrozenxid  (Jerry Sievers <gsievers19@comcast.net>)
Список pgsql-performance
Дерево обсуждения
Hey,
I'm trying to help a guy that is using pg9.6 but I'm not so familiar with the error message : 
ERROR:  found xmin 16804535 from before relfrozenxid 90126924
CONTEXT:  automatic vacuum of table db1.public.table_1"


It seems that the error has started appearing two weeks ago.  Data that I collected : 

-all the autovacuum params are set to default

-SELECT relname, age(relfrozenxid) as xid_age, 
    pg_size_pretty(pg_table_size(oid)) as table_size
FROM pg_class 
WHERE relkind = 'r' and pg_table_size(oid) > 1073741824
ORDER BY age(relfrozenxid) DESC LIMIT 4;
            relname            |  xid_age  | table_size 
-------------------------------+-----------+------------
  table_1            | 180850538 | 10 GB
table_2       | 163557812 | 10 GB
table_3 | 143732477 | 1270 MB
table_4       |  70464685 | 3376 MB

pg_controldata : 
Latest checkpoint's NextXID:          0:270977386
Latest checkpoint's NextOID:          25567991
Latest checkpoint's NextMultiXactId:  1079168
Latest checkpoint's NextMultiOffset:  68355
Latest checkpoint's oldestXID:        77980003
Latest checkpoint's oldestXID's DB:   16403
Latest checkpoint's oldestActiveXID:  0
Latest checkpoint's oldestMultiXid:   1047846
Latest checkpoint's oldestMulti's DB: 16403
Latest checkpoint's oldestCommitTsXid:0
Latest checkpoint's newestCommitTsXid:0

It seems that the autovacuum cant vacuum table_1 and it has alot of dead_tuples. Moreover, it seems that the indexes are bloated.

schemanamerelnamen_tup_updn_tup_deln_tup_hot_updn_live_tupn_dead_tupn_mod_since_analyzelast_vacuumlast_autovacuumlast_analyze
publictable_105422370038222210958292310760701

I tried to vacuum the table (full,freeze) but it didnt help.
I read about the wrap that can happen but to be honest I'm not sure that I understood id. 
What can I do to vacuum the table ? Can some one explain the logic behind the error message ?

Thanks.

В списке pgsql-performance по дате отправления:

Предыдущее
От: Jim Finnerty
Дата:
Сообщение: Re: SELECT performance drop
Следующее
От: Jan Nielsen
Дата:
Сообщение: Re: SELECT performance drop