Forum Discussion
NimbostratusActive/Standby Trunk Port Migration
I need to migrate the internal based trunks on the Ha pair from 1Gbe copper to 10Gbe fiber and had a few questions.
Does the physical network connections have to be the same on both cluster members? The reason why I am asking is I currently have an active/standby clusters with the inside vlan consisting of a (4) 1GBE copper ports on both the active and the standby. I need to migrate off the (4) member 1Gbe copper based trunk to a (2) member 10Gbe fiber port trunk. My plan was to perform the migration on the standby unit first and then fail from the active to the standby and do the same thing on the previously active node which will be in standby after the forced failure.
I was just curious if the active is still using the 1Gbe based trunks and the standby using 10Gbe trunks, if it will cause a problem when performing the fail over. Trying to migrate the trunks with minimal impact to the traffic, since the boxes are in production and can't be taken out to perform the migration.
Any feedback or experiences on the topic would be greatly appreciated.
6 Replies
Michael_YatesHi Silvajai,
Your methodology should work fine. The network configuration for each device (including trunk and interface assignments) are device specific.
As long as the Standby device can communicate on those VLAN's you should not have any problems (which you will know pretty quickly if your Health Monitors start working).
silvajai_01_131That's what I thought. Thanks for the confirmation!
Michael_Yates_6Cirrus
Hi Silvajai,
Your methodology should work fine. The network configuration for each device (including trunk and interface assignments) are device specific.
As long as the Standby device can communicate on those VLAN's you should not have any problems (which you will know pretty quickly if your Health Monitors start working).
- silvajai_01_131That's what I thought. Thanks for the confirmation!
hoolioCirrostratus
From an internal poll :)
Adding the vlans to both trunks at the same time causes a spanning tree convergence or an L2 loop--both of which are bad. Using a TMSH transaction to remove the tags from trunk1g, and apply the tags to trunk10g is fast, but not so fast that you don’t drop packets (monitors, however, won’t likely fail, fwiw).
Doing the standby box, fail over, and then do the original box’ is the best practice.
Aaron
- silvajai_01_131
My plan is to perform the migration on the standby unit first. I plan on removing the existing trunk that contains the tagged vlans and then replace it with new 10gbe based trunk supporting the same tagged vlans. Once I verify that the health monitors are able to reach the support nodes, I plan on failing over from the active node to the standby and doing the same thing on the previously active node. This should avoid the stp looping issue that you described.
