02-17-2016 10:07 PM - edited 03-08-2019 04:37 AM
Dear Team,
Customer had a concern that he want do upgrade the L3 device with minimum downtime.
Current version running is 12.2 and upgrading to new version 15.0.2.
He is concern of zero downtime with respect to access Switch.
He has 2960 sw at a access later and has etherchannel configured towards L3 (i.e. one link goes to SW1 and other link goes to SW2). This L3 switches are stacked.
Now he is planning to load the image in master and slave.
He will reboot only the master SW1 and now the stack will break and traffic of 2960 will go from Sw2.
After SW1 comes up they will reboot SW2 and now the traffic of 2960 will go from SW1.
After both are on the same version he will form the stack of both the Switches so that user dont face any issue or we can upgrade with minimum downtime.
Pls let me know whether I can go with this procedure.
Pls find the attached for the network diagram.
Thanks in advance
Regards,
Vikas
Solved! Go to Solution.
02-17-2016 10:20 PM
Make sure you go to the gold star release 15.0.2-SE9.
02-17-2016 10:16 PM
This is not going to work. When you reboot one member; when it restarts and if it is running a different version than the master (and it will be) it will then change its software version to that of the master.
If the client wants no-interruption then they should have bought a higher end switch that supports ISSU (in-service software upgrade) like a 4507 with dual supervisors.
There is no way to avoid the downtime doing the software upgrade in this setup. I would plan for a 15 to 20 minute outage, especially since you are doing a massive version jump. It is highly likely the switches will do double reboots to upgrade embedded controllers.
02-17-2016 10:23 PM
Hello Philip,
Thanks for your immediate response.
I am planning to do so..pls let me know whether it is possible.
We can disable the stack ports break the stack, boot the master, then after master comes up then we will boot the slave and later again enable the stack port.
Pls let me if there is any other way in which i can reduce the downtime to 5 -10 mins only
Regards,
Vikas
02-17-2016 10:26 PM
If you break the stack and you have any cross-chassis Etherchannels then they'll be screwed. Also the layer 2 forwarding path will be broken between the 3750's. I think it will go badly.
It *might* complete in 10 minutes. I'll give you a 75% chance of it finishing in 10 minutes. It is a big upgrade you are doing.
02-17-2016 10:30 PM
I have recently upgraded other devices with the same model it took me 30-40 mins to upgrade and join the stack back.
So planning to do this way.
Regards,
Vikas.
02-17-2016 10:33 PM
Alas you know the answer to tell the client ...
02-17-2016 10:51 PM
Yes I already told him but the problem is customer dont have a down time of 30 mins. He is ready to give a max of 5-10 min.
that where the problem is...
Is there any other way like rebooting the slave just before stackwise protocol initiation happen or anything like that so that I can alteast reduce the downtime.
Regards,
Vikas
02-18-2016 10:32 AM
The only other options are more extreme.
02-17-2016 10:20 PM
Make sure you go to the gold star release 15.0.2-SE9.
02-17-2016 10:26 PM
Hello Philip,
We are upgrading to this version only.
Thanks,,
Regards,
Vikas
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide