Am 29.11.2011 16:46, schrieb Phoenix Kiula:
About 5% of the times (in situations of high traffic), this is not returning a value in my PHP code. Because it's not found, the code tries to INSERT a new record and there's a duplicate key error, which is in the logs. The traffic to the site is much higher than the number of these entries in my log, which means it's only happening sometimes--my guess is for 5% of all queries, which is still quite significant (about 60,000 a day).
As the two statements aren't atomic (i.e., after the select, how do you make sure that no other process accessing the DB has a chance to select/insert, before the [delayed] insert of the primary process gets executed)? This is a common form of race-condition.
In the spirit of "it's easier to ask forgiveness than permission", just do the insert and evaluate the result (i.e., whether a record was actually inserted) to get at the same information and the same effect as with the two statements you're currently executing.
-- --- Heiko. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
