-----Original Message-----
From: Guillaume Radde [mailto:guillaume.radde@redhat.com]
Sent: Sunday, March 25, 2012 5:28 AM
To: Tripp, Donald (HAL)
Cc: HAL DistList: IS UNIX; Flach, Cynthia (HAL); Ruckman, Maurice (HAL); jlindesm@redhat.com; Augustine, Anila (HAL); Hofsetz, Therron (HAL); Fillman, Eric (HAL); Thompson, Sonya (HAL); Phatak, Sheetal (HAL); Bojja, Sridhar (HAL Contractor)
Subject: Re: VMware status
Thanks Don,
If the servers need reboot while live, we can also disable half of the cluster through mod_jk without loosing session or outage. The algorithm will look like this:
a) On halprdjbs01 and halprdjbs02
-Disable halprdjbs01 and halprdjbs02 in halprdweb01 through jkstatus -Disable halprdjbs01 and halprdjbs02 in halprdweb02 through jkstatus -Wait for existing sessions to expire in haljbs01 and haljbs02 (sessions draining, can take several hours depending on usage) -Reboot halprdjbs01 and halprdjbs02 -Re enable halprdjbs01 and halprdjbs02 in halprdweb01 through jkstatus -Re enable halprdjbs01 and halprdjbs02 in halprdweb02 through jkstatus
b) Same on halprdjbs03 and halprdjbs04
-Disable halprdjbs03 and halprdjbs04 in halprdweb01 through jkstatus -Disable halprdjbs03 and halprdjbs04 in halprdweb02 through jkstatus -Wait for existing sessions to expire in halprdjbs03 and haljbs04 (sessions draining) -Reboot halprdjbs03 and halprdjbs04 -Re enable halprdjbs03 and halprdjbs04 in halprdweb01 through jkstatus -Re enable halprdjbs03 and halprdjbs04 in halprdweb02 through jkstatus
