Detecting corrupt table

Поиск
Список
Период
Сортировка
От David Larochelle
Тема Detecting corrupt table
Дата
Msg-id CABipw1mtbhkYJZwo2QZ7HLc3+YadW2XA_Jc5XD6=-8dKh0Uj4w@mail.gmail.com
обсуждение исходный текст
Ответы Re: Detecting corrupt table
Список pgsql-general
Дерево обсуждения
Our database has some corrupt tables and I'm trying to figure out what data can be salvaged and what needs to be restored from backup or regenerated.

Initially I tried running select count(*) on all user tables. While this did detect some corrupt tables, it missed others. For example, I was able to run count(*) on a table but then got an error while trying to back it up.

pg_dump: Error message from server: ERROR:  missing chunk number 0 for toast value 368243665 in pg_toast_284730161
pg_dump: The command was: COPY public.stories (stories_id, media_id, url, guid, title, description, publish_date, collect_date, story_texts_id, full_text_rss) TO stdout;


Is there a simple way to determine which parts of the database are corrupt? I'm currently running a script to back up each table individually using something like the following:

psql -c "select tablename from pg_tables where tableowner = 'db_user' ORDER by tablename " | tail -n +3 | head -n -2 | xargs -n 1  -i pg_dump --verbose --table={} --file={}_.dump

But I'm worried that this approach will also miss database corruption and was wondering if anyone has other suggestions.

Thanks,


David


В списке pgsql-general по дате отправления: