Wondering if we had a situation wherein we were in the middle of ontap upgrade ..and it was decided due to some issue NOT to perform the second node TO/GB.
How long can these two nodes be running different versions until some issue would surface?Note that we are very well aware that both the heads should be at different version for minimum amount of time,but how long can this time be ?
We had an issue with the cisco switch ports during the first node TO/GB and suspect that this is a network side issue hence this kind of approach was being thought?
well you could run it a long time but in the event of a system failover it probably will not takeover due to the mismatch even if it is a minor upgrade release.
Last time this happened to me I decided to revert, then at a later time attempted again the upgrade. It is important to do a full check before upgrading as best practices to avoid this situation but it happens especially if certain checks are out of your control.
I would take a look at the upgrade/revert guide for what version of ONTAP you were upgrading.
Thanks ..what you say about an accidental takeover not working until both the versions are same makes sense.
However in our case it was not a problem of improper pre-checks.After the first TO/GB we noticed that due to LACP configuration and a bug (Bug id # 337379) the network ports wne went offline/unresponsive causing all the vfilers to be not reachable causing complete outage.A shut/noshut was needed on the network ports to finally ensure that the vfilers were reachable (i.e the VIF was up and running again).So we have both the nodes running different ontap versions as of now.
We are checking what caused the network ports to go offline (from network team side) before performing the second TO/GB.
Our nest step is - to find the RCA for port offline and ensure that everything is fine on the network switches (LACP,trunking etc ) before performing the second node TO/GB .
So wondering how did you finally finish your upgrade?