Forum Discussion
Gerald_Meese
Cirrus
CirrusCopy over self IPs from several partitions
Hello,
I'm preparing the migration of an active/standby cluster of 2 LTM devices to new virtual devices.
Per F5 support case I cannot use an UCS migration due to different trunk settings on both sides.
So I'm using the "merge file" feature as described in K81271448. We have 3 partitions so I exported the SelfIP's using 3 commands :
tmsh -c "cd /Common; list net self" > /var/local/scf/Common-selfip.txt
tmsh -c "cd /Partition1; list net self" > /var/local/scf/Partition1-selfip.txt
tmsh -c "cd /Partition2; list net self" > /var/local/scf/Partition2-selfip.txt
For the import I'm using 3 commands :
tmsh -c "cd /Common; load /sys config merge file Common-selfip.txt"
tmsh -c "cd /Partition1; load /sys config merge file Partition1-selfip.txt"
tmsh -c "cd /Partition2; load /sys config merge file Partition2-selfip.txt"
Import of /Common-selfip.txt went well, but for Partition1 and Partition2 I get this error :
01070978:3: The vlan (/Common/xxxxx) for the specified self IP (10.160.10.118) must be one of the vlans in the associated route domain (/Common/0). For example: 192.168.0.1%1234 for self IP in route-domain 1234.
I edited the Partition1-selfip.txt file and modified the IP to 10.160.10.118%1, the error for that entry has disappeared... but the SelfIP is not created (in any of the partitions) š
Could anyone please point me to the right set of commands to export/import all Self IP's in their respective partition ? That would be highly appreciated š
Have a good day,
Gerald
I wonder if this is as simple as not setting a default route domain for your other partitions? Try to set that before the import, rather than modifying the IPs to put the route domain in with the %xxx syntax. You have to first define the route domain object (net route-domain), then set the partition to it.
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# create net route-domain rd_foo id 1234 vlans replace-all-with { con1 }
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# list auth partition foo all-properties
auth partition foo {
default-route-domain 0
description none
}
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# modify auth partition foo default-route-domain 1234
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# list auth partition foo all-properties
auth partition foo {
default-route-domain 1234
description none
}
- Lucas_Thompson
Employee
I wonder if this is as simple as not setting a default route domain for your other partitions? Try to set that before the import, rather than modifying the IPs to put the route domain in with the %xxx syntax. You have to first define the route domain object (net route-domain), then set the partition to it.
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# create net route-domain rd_foo id 1234 vlans replace-all-with { con1 }
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# list auth partition foo all-properties
auth partition foo {
default-route-domain 0
description none
}
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# modify auth partition foo default-route-domain 1234
admin@(west)(cfg-sync Standalone)(ModuleNotLicensed::Active)(/Common)(tmos)# list auth partition foo all-properties
auth partition foo {
default-route-domain 1234
description none
}
Gerald_MeeseHi Lucas,
Thanks for your answer ! Actually the 3 partitions have their own route domain :
auth partition Common {
default-route-domain 0
description "Repository for system objects and shared objects."
}auth partition Partition1 {
default-route-domain 1
description none
}auth partition Partition2 {
default-route-domain 2
description none
}What is really strange is this : āI edited the Partition1-selfip.txt file and modified the IP to 10.160.10.118%1, the error for that entry has disappeared... but the SelfIP is not created (in any of the partitions) šā.
Usually when preparing migrations I add SelfIPās manually but this customer has +/- 150 (!) of them across 3 partitions ā¹
Have a good day,
Gerald
- Gerald_Meese
I found the issue š when I use a concatenated command to move to Partition1 and run the import, it fails (silently) and stays in partition /Common :
tmsh -c "cd /Partition1; load /sys config merge file Partition1-selfip.txt verify"When I do the same using 2 commands in sequence... it works like a charm š
Thanks for your help!!
Gerald
- zamroni777
Nacreous
During Ericsson POC in 2011 with F5 presales, I was able to pair Viprion with VE.
Even persistence records were synchronized.
So, I suggest you to try:
1. break current pairing then shutdown the standby unit.
2. pair the active unit with one of the new HW.
3. switchover the new HW to active node
4. break pairing then shutdown the standbay unit (old HW)
5. pair with second new HW- Gerald_Meese
Thanks for your reply ! The steps you listed are actually my plan š but before to do that I have to configure the network objects on the new servers. Given the large amount of objects I tried to automate that as much as I could but I'm facing some issues.
- zamroni777
because old standby HW is shutted down, you can use the old standby's physical links and self & mgmt IP configuration on the new HW.
After that, pair this new HW as standby of old active HW.
