By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
464,302 Members | 1,163 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 464,302 IT Pros & Developers. It's quick & easy.

WAL question

P: n/a
hi, all,

I have just read Postgresql-7.3.4 documents about WAL (write ahead
logging). But there are still something unclear to me. I'm not sure my
understanding is right. Please let me know if there is anything wrong. Thank
you all in advance.

WAL means any changes to data pages shouldn't be written to disk before
the its log record have been written to log file. If log records is written
to disk everytime the commit is performed, there is no uncommitted changes
written to database. So there is no undo action needed during recovery time.
Another alternative is flushing log buffer once it is full. In this case,
un-committed changes are possible written to databases. So undo action is
inevitable for after-crash recovery.

In postgresql-7.3.4, UNDO operation is not implemented because of a
permanent pg_clog file to hold the status of transactions. My question is
that how this trick works precisely ?

Jack Dull

Nov 11 '05 #1
Share this question for a faster answer!
Share on Google+

This discussion thread is closed

Replies have been disabled for this discussion.