USN-2124-1 (OpenJDK-6) has regressed..?

[William Scott Lockwood III]

On Tue, Mar 4, 2014 at 3:27 PM, James Green <james.mk.green at gmail.com> wrote:
> William,
> The application concerned has been chugging away without major changes for over six years. The last MySQL update received was late January. Soon as the java hot fix was installed Fri morning we are observing chaos.
> We have cases where the app lasts 48hrs, others where it lasts less than a minute on a lightly loaded system. Some of our machines have observed a decreased rate of use in recent years as traffic has migrated toward newer instances of hardware so we aren't experiencing sudden growth that we need to adjust connection limits for.
> The app chugs away unchanged across the fleet. We have one box where oracle jdk 6 is installed and the app there hasn't crashed at all. On all the open jdk 6 boxes we get crashes.
> The MySQL console shows literally dozens of connections in a sleep state when things go wrong. On our larger db instances there are hundreds. Each starving front ends.
> I am going to try and file a bug I think as the Ubuntu package is the only change we can see.
> James
> Sent from my iPad


Interesting. Makes sense I suppose. When this happened to us, it was
the database that was really the issue (and had been running in a
stable state for years upon years as well) but, as complex as software
is, it's not always the same thing that causes a problem.

Have you tried reverting to the previous java package? Lastly, it's an
ugly hack, but have you considered running a job out of cron that
looks for, and kills sleeping connections? Ugly i know, but if that
java update was fixing security issues...

--
W. Scott Lockwood III
"There are four boxes to be used in defense of liberty:  soap, ballot,
jury, and ammo. Please use in that order." -Ed Howdershelt (Author)