cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1460
Views
0
Helpful
4
Replies

N5K UPGRDAE

Hi Pros,

 

First of all thank y'all for the help. I am in the preparation to upgrade a pair of N5K in the upcoming weekend, they've been in prod since 2012 and never been upgraded. Now Cisco has sent me an email about the NX-OS running on these switches won't be supported at the end of June. So here i am i have to upgrade... I have read the upgrade guide a few times and plan to re-read at least once more, i have watched some videos. It's all seems easy!... but still  nervous about upgrade in prod. Do not have a lab to try things out, so i have to go blind!

I hope some of you had done this  before and can give me a few pointers...a bit of guidance.

 

Here is my upgrade path: 5.1(3)N2(1) >> 5.2(1)N1(9b) >> 7.0(8)N1(1) >> 7.1(4)N1(1) >> 7.3(3)N1(1)

We're currently running 5.1(3)N1(1a) in vPC mode, default zoning and EMC SAN and UCS connected to them, meant the FI switches , a pair of 6248 UP. I am the only network guy here, and i have a 4 hours window for the both switches.

I have downloaded the firmware already, so  have I to upgrade a kickstart for each OS? for instance kicksart for 5.1 and kickstar for 5.2 and so one and so forth.

Now as mentioned below, i am using zone in the N5Ks and UCS side has been configured with default zoning only. If i lost my zone config, how do i get them back? Just upload a new config to the SW? And final test to make sure it all works? I have read this blog which also pointed Zoning issue after upgrade: https://community.cisco.com/t5/networking-documents/n5k-7-0-1-n1-1-upgrade-from-5-0-5-2/ta-p/3136581

And should i wait till my final upgrade(7.3(3)N1(1)) to do my testing? Or should i test as i go alone?

 

thanks,

4 Replies 4

rlaneyjr22
Level 1
Level 1

I have upgraded my share of Nexus5k cores with 4hour windows for each switch.  The upgrade path you have chosen will most likely go over your 4hr window.  You are already looking at 4 complete reboots with at least one full power disconnect to upgrade the hardware revision.  If you are running FCOE over FEX ports you will want to read below.  This was a big deal for many sites since this caused every server VM to loose it's hard drive completely.  Try unplugging the hard drive to a DB server and you will see what I mean.

This issue is related to a behavior change that was incorporated starting 7.0(6) and 7.1(1) whereby an implicit class-fcoe to cos3 mapping was created (a resolution to bug ID CSCuq86032 as documented in the 7.x release notes: https://www.cisco.com/c/en/us/td/docs/switches/datacenter/nexus5500/sw/release/notes/7x/Nexus5500_Release_Notes_7x.html). The purpose was to prevent and error out match overlaps when fcoe was in use but appears to have had the side effect of pushing that implicit match even when fcoe was not in use and was later remedied and provided a work around is provided and tracked through bug-ID CSCuu06699:
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCuu06699/?referring_site=ss
********************************************************************************
Conditions:
cos 3 traffic match with any user defined policy will not work as cos 3 is reserved for fcoe

********************************************************************************

Often I found it easier and quicker to take the hit of lost config and avoid the super-long nights by doing a one-step upgrade (where possible).  Then using my always prepared back-door console to repair the config.

 

Good Luck!

Thanks mate for the reply. Really appreciated it.

As i am going through the release note, i realized that it is going to be a long night... Thanks for pointing it out and confirming it. As i have to do the two switches in the same window, i will try to top at 7.1(4)N1(1) and will plan for another window to complete the upgrade to 7.3(3)N1(1).

 

You mentioned two things that i think  won't affect us much, fingers crossed. First all our servers are B-Series with internal HD, no boot from SAN. However, the Switches do have the zoning configuration, so the servers can access shared on the SAN. If the zoning config wipes out, i will definitely have major issues.

Second we have fcoe feature enable but we use pure fc to the SAN and the 6248UP FI connected to the UCS chassis.

 

Thanks,

 

How was your upgrade? Any challenge that you encountered and how do you solve it?

Upgrade went well.
Except in the last image i had issues with one kind of SFP fiber module. The firmware did not support it, i had to back out. Ordered new module, and upgrade to 7.3(3)N1.
Good luck!

Review Cisco Networking for a $25 gift card