On 14 September 2010 10:00, JL <[email protected]> wrote: >> Then might I recommend an Occam's Roazor approach to this? Compile a >> number of kernels from the one which worked up to the one which doesn't. >> Reboot the backup; test; if it doesn't happen then you know the change >> cam later than the kernel you're running. > I intend to do that.
The results: 2.6.27.45 - OK 2.6.27.53 - OK 2.6.28 - Fail 2.6.28.10 - Fail 2.6.29.6 w/ Classic RCU - Fail 2.6.29.6 w/ Tree RCU - Fail 2.6.32.8 - Fail 2.6.35.4 - Fail I'm convinced the problem was introduced in 2.6.28. I still believe this is somehow related to the insertion of connection entries received from the master into some in-memory table on the backup. -- Jarrod Lowe _______________________________________________ Please read the documentation before posting - it's available at: http://www.linuxvirtualserver.org/ LinuxVirtualServer.org mailing list - [email protected] Send requests to [email protected] or go to http://lists.graemef.net/mailman/listinfo/lvs-users
