Unsolved
This post is more than 5 years old
24 Posts
0
3305
Datamover reboot order post OE upgrade?
Hi,
i'm doing an upgrade on a vnx5200 with two datamovers and have opted to reboot the datamovers post upgrade. This is a production box with NFS presented to ESXi servers. I have the commands to halt and reboot the DMs but i wanted to know if there is a recommendation on the order.
Currently i am thinking I would do the following:
1. halt/reboot server3
2. failover to server3
3. halt/reboot server2
4. failback to server2
Thanks in advance
Gabe
dynamox
2 Intern
2 Intern
•
20.4K Posts
1
June 12th, 2014 07:00
during a normal upgrade process, support upgrades and reboot the standby datamover first, then they upgrade the primary one. I have always used failover/failback ..did not have to halt anything.
Rainer_EMC
8.6K Posts
0
June 12th, 2014 07:00
Its your choice – a reboot takes a bit longer than a failover but then you don’t need a failback.
So it’s a choice of one slightly longer downtime vs. two smaller ones
For example from recent testing with a small sized VNX 5500 the protocol downtime for a warm reboot is 30s vs. for failover/failback it is 20s
In my experience a properly setup VMware NFS data store will have no problems with up to 180 seconds
gheinisch
24 Posts
0
June 12th, 2014 07:00
Okay, so failover and failback will reload the code and I don’t need to do a reboot. That is what you are saying correct?
Gabe Heinisch
919-280-7112
gheinisch
24 Posts
0
June 12th, 2014 07:00
okay, thanks. I will do the following:
#upgrade standby DM (server_3)
server_cpu server_3 -reboot -monitor now
#failover to standby DM
server_standby server_2 -activate mover
#upgrade primary DM
server_cpu server_3.server_2.failed -reboot -monitor now
#failback to primary
server_standby server_2 -restore mover
gheinisch
24 Posts
0
June 12th, 2014 09:00
Great, it’s all done and no issues at all with the NFS. the longest cutover was from secondary back to primary. it took a total of 1 minute 7 seconds.
many thanks
Gabe
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
June 12th, 2014 09:00
Correct, when you failover to standby primary gets rebooted.
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
June 12th, 2014 09:00
How many file systems do you have ?
Rainer_EMC
8.6K Posts
0
June 16th, 2014 02:00
Just curious – how did you measure the „1 minute and 7 seconds“ ?
gheinisch
24 Posts
1
June 16th, 2014 05:00
sorry for not responding earlier to your other inquiry. the customer has 4 NFS fs and we started a ping session before we swapped the movers. so at the very least, he measured a minute seven from the moment we lost ping until we had ping back…not exactly stating the DM was back up but to some extent.
Rainer_EMC
8.6K Posts
0
June 16th, 2014 05:00
Ok – it does sound a bit long for just 4 fs though