Forum Discussion
Fletcher_Cocquy
Nimbostratus
Jun 17, 2011Upgrading LTM HA Pair - disable network failover?
Hi All -
having successfully upgraded our standalone LTM unit from 9.3.1->9.4.8->10.2.1, I'm trying to find the recommended practice for upgrading our production HA pair - does this make sense:
1) upgrade Standby unit
1a) backup config
1b) disable network failover ? (to prevent this unit from trying to become ACTIVE during upgrades/reboots)
1c) upgrade standby to 9.4.8: im local-install-9.4.8.355.0.im
1d) standby reboots
1e) re-activate license
1f) upgrade standby to 10.2.1: image2disk --instslot=HD1.1 --format=volumes BIGIP-10.2.1.297.0.iso
1g) standby reboots
1h) login to standby, check config looks all there
1i) re-enable network failover?
1j) swap active/standby preference on both active and standby units, and force active unit to standby
2) repeat steps for the remaining (was active - now standby) unit
I want to be ABSOLUTELY sure I don't get an active/active race during the upgrades - should I uncheck the Network Failover Enabled box?
Or will the active/standby preference be sufficient?
thanks
4 Replies
- nitass
Employee
why do u need to upgrade to 9.4.8 before going to 10.2.1?
1b, if network failover is disabled, both unit will go active. i think u should disconnect all network connectivity.
sol11215: Upgrading an active/standby BIG-IP high availability pair
http://support.f5.com/kb/en-us/solutions/public/11000/200/sol11215.html
1f, image2disk --format=volumes BIGIP-10.2.1.297.0.iso
sol12561: Change in Behavior: The image2disk utility --instslot and --format options are now mutually exclusive
http://support.f5.com/kb/en-us/solutions/public/12000/500/sol12561.html
there are some change behavior in v10.x. these are what u should take note.
sol9947: Change in Behavior: The Peer Management Address setting is required for BIG-IP version 10.x systems configured for network failover
http://support.f5.com/kb/en-us/solutions/public/9000/900/sol9947.html
sol10655: Change in Behavior: CR/LF characters appended to the HTTP monitor Send string
http://support.f5.com/kb/en-us/solutions/public/10000/600/sol10655.html
finally, if there is any issue in loading config after upgrading, try b load and it should show what is wrong.
hth - Fletcher_Cocquy
Nimbostratus
9.4.8 is the support recommended path to 10.x:
"The upgrade path for LTM, in order to correctly roll forward your config, is ; go to version 9.4.8, confirm config is good, then do a v10 upgrade."
thanks for the link - I can't believe the "search F5" field does not seem to access those knowledgebase articles!
http://support.f5.com/kb/en-us/solutions/public/11000/200/sol11215.html?sr=15041414network_failover_mgmt says:
Temporarily set the Redundancy State Preference setting of both units to None. This action ensures that the BIG-IP systems will not swap roles unnecessarily during the procedure.
The doc goes on to say disconnect the unit you are upgrading from the networks except for mgmt (for heartbeats)
If the new & old High availability protocols are compatible then you are in luck otherwise you need downtime to complete the upgrade of the pair.
So is there non-disruptive path from my current 9.3.1 to latest 10.2.1 without downtime? (are the High availability protocols compatible?)
thanks - nitass
Employee
i don't think network failover protocol is compatible even port is identical. i mean it would have some downtime.
sol9057: Service port and protocol used for BIG-IP network failover
http://support.f5.com/kb/en-us/solutions/public/9000/000/sol9057.html - nathe
Cirrocumulus
I'm agreed with nitass re network failover protocols - even if the two versions use the same protocol / port i would play safe and not bank on them working together as you'd hope. Doesn't one of the referenced docs give this warning / advice? I think it does. Anyway, for a non-downtime scenario I used serial failover, rather than network failover. Then the Active would remain Active whilst you upgrade / reboot the Standby. It made my life a lot easier when I upgraded my HA environment. If I remember correctly I purchased one from f5 - and not expensive at all.
Re Redundancy State Preference - I'd always have this set as None. I inherited them configured with a preference but this only caused unnecessary failovers so changed it to None. For example, an Active reboots for some reason, Standby now becomes Active. However, once the original Active LTM is back online it then is made Active again. Just my two pence here.
N
Help guide the future of your DevCentral Community!
What tools do you use to collaborate? (1min - anonymous)Recent Discussions
Related Content
DevCentral Quicklinks
* Getting Started on DevCentral
* Community Guidelines
* Community Terms of Use / EULA
* Community Ranking Explained
* Community Resources
* Contact the DevCentral Team
* Update MFA on account.f5.com
Discover DevCentral Connects