On Jun 27, 2013, at 8:47 PM, Paul M Foster <paulf@xxxxxxxxxxxxxxxxx> wrote: > On Thu, Jun 27, 2013 at 11:47:28PM +0200, adriano wrote: > >> holes in sequence of auto increment happen using transaction for >> insert new record but you don't commit transaction itself >> it seems that the autoincrement is incremented anyway >> at least this is my case. > > I think what Tedd was referring to was something else. The "hole" was > quite large. I've seen this behavior myself, in PostgreSQL. From one > transaction to the next, there were over 10,000 skipped numbers, and > only me and my wife on the system. Some sort of bug, like a spinlock > that wasn't interrupted the way it should have been. I remember the > system taking forever to calm down before it gave the next transaction a > number way forward of the last one. I waited in front of my browser for > quite some time. But I couldn't explain why. > > Paul Yes, it was something like what Paul said -- it was not a transaction skip. I don't know what to think about it -- no explanation. But, the problem suddenly vanished -- very strange. Cheers, tedd _____________________ tedd.sperling@xxxxxxxxx http://sperling.com -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
