Forum Discussion
nanukanu_263047
Nimbostratus
NimbostratusVCMP Hosts and Guests
Hi community!!
Recently I asked refered to HA betwen Guests, and Hosts, about sync... And actually due to the answer recieved I'm realizing the HA process between Guests (Active / Passive), and ...
Hi,
You have to create the vlans again. But as long as you already configured those vlans on the first host, you can copy the vlan config in bigip_base.conf file from the first host and paste this configuration to the second host on the same file.
You should copy all vlan information like below :
net vlan /Common/vlan_external { interfaces { 1.2 { } } tag 4093 }Once copied to the second host, you save the bigip_base.conf file and execute a
to apply your changes.tmsh load sys configbigip_base.conf is located under /config
Yann_Desmarest
Cirrus
CirrusHi,
You have to create the vlans again. But as long as you already configured those vlans on the first host, you can copy the vlan config in bigip_base.conf file from the first host and paste this configuration to the second host on the same file.
You should copy all vlan information like below :
net vlan /Common/vlan_external {
interfaces {
1.2 { }
}
tag 4093
}
Once copied to the second host, you save the bigip_base.conf file and execute a
tmsh load sys configbigip_base.conf is located under /config
nanukanu_263047Perfect! I supose it, so, its the correct way of proceed, correct? And related with this -I only see 2 Interfaces on Guests (on Host aggregate of 4) it's correct? -I can use the same IP on ACTIVE/STANDBY vlan Interface? or its imperative play with floating IP? Thanks!- Yann_DesmarestThe procedure above works for vlan replication on the host only, I already noticed that the guest is not displaying correct interfaces, speed, etc. but the important is that you have your vlans propagated on the guest. Regarding the Active/Standby, you will need to share a floating Self IP for outgoing traffic to peers (snat automap feature). You cannot assign the same physical Self IP to multiples Guests
- Yann_DesmarestThe procedure above works for vlan replication on the host only, I already noticed that the guest is not displaying correct interfaces, speed, etc. but the important is that you have your vlans propagated on the guest. Regarding the Active/Standby, you will need to share a floating Self IP for outgoing traffic to peers (snat automap feature). You cannot assign the same physical Self IP to multiples Guests
- Yann_DesmarestOne more thing, if you are running vcmp on VIPRION, there is some limitations. You must define each guest of a same device group in the same blade number (even on different viprion chassis), and you should not have guests that run accross multiple blades
- tatmotiv
Cirrostratus
@Yann Desmarest: "You must define each guest of a same device group in the same blade number (even on different viprion chassis), and you should not have guests that run accross multiple blades" Do you have any source for this statement? - Yann_DesmarestHi, I have several painful hours and case opened with F5 last year for those topics. They confirmed that, in an HA configuration, I should not run a guest on VIPRION 1 Blade 1 and the other guest on VIPRION 2 Blade 2. We get many sync and mirroring issues cause of this.
- Yann_DesmarestFor clustering, you should configure a guest to run on multiple blades or in HA with another guest on another chassis. Running a cluster of guest on the same chassis and blades is not supported.
- Yann_DesmarestSome info here on the "High availability considerations" section : https://support.f5.com/kb/en-us/solutions/public/15000/900/sol15930.html?sr=54484643
- tatmotivThanks for sharing. I was more concerned about the "you should not have guests that run accross multiple blades" parts though... The SOL you are referring to puts it as follows: "HA device groups are configured on the vCMP guests. Members of a Sync-Failover device group should each run on a separate chassis, and each peer in the device group should have the same number of slots on each chassis". So - as I understand it (and was also my understanding before reading this post), it is perfectly safe to have an HA group of vCMP guests using more than one blade each, as long as the number of allowed slots is the same across the vCMP hosts (haven't tried this yet, though). Do you disagree?
- Yann_DesmarestAs I told before, you have the following options : 1) you can configure a guest to spread accross multiple blades on the same chassis. 2) you can configure two guest in HA, but they must be single blade, same blade number but different chassis. When configuring two guests in HA and each guest spread accross multiples blades, we faced replication issues between blade for a guest. It's even worst if you run other modules like ASM, APM. Check your configuration with F5 to make sure that your scenarios are supported