This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
A little bit more than once daily nb.org site isn't accessible because of this DB error: Problem clening up old session data: SQLSTATE[HY000] [2003] Can't connect to MySQL server on 'nbxdbsha' (4) I asked Honza P. for information how often is the site unavailable and it was eight times in last week (it means 10/5 - 17/5). It's happening in various times - yesterday (17/5) it was about 9:30am, on Friday (14/5) 3:28pm PRG time. Every outage takes about 5 minutes.
corresponding kenai issue for that: http://kenai.com/jira/browse/KENAI-1995
update: for better tracking, the PHPstack is now logging all DB exceptions to the logfile. I provided the kenai team with the file location.. Now we should know real scope of those outages... honza
I'm attaching data of incidents in 21/5-25/5 (till was done logging of these issues) which were detected and sent by email from @pingability site: 21/5 - 1 the same incident, 22/5 - 1, 23/5 - 1, 24/5 - 3, 25/5 - 4
Created attachment 99498 [details] DB exceptions log from May 25, 26 This is the log of DB exceptions. Showing all requests which failed due to some DB issue
Honza, would you be able to add some more logging on the exception? I would be interested to know why it failed:-( Some catch blocks like mentioned in: http://zend-framework-community.634137.n4.nabble.com/Suddenly-two-of-my-sites-using-Zend-Db-stop-working-td673581.html or something, with some info about the found exception: Network, too many clients, ..., ??? Thanks, -- Marco
Hi Marco, I added better catch block as suggested on Zend_Db documentation. Let's see the result tomorrow. honza
Created attachment 99667 [details] DB exceptions from 2010-05-28 to 31 DB exceptions from 2010-05-28 to 31, after I changed the catch block and switched from DB_Pdo to Mysqli DB adapter
On 16/06 by NB6.9 release (higher load) was the site unavailable at least 8 times again. Did the improved logging help for better identification of the issue?
Since Lea doesn't work for Oracle or on Kenai anymore, we might want to reassign this to an actual person that can and might do something to resolve the problem... I do not know who that person would be... If I did, these issues would be assigned to them when I committed this comment.
Works for now - so closing
Agree, looks that it's already fixed.