I just successfully upgraded 5 instances of NSX 3.2.2. to 4.1.2 (using VMware-NSX-upgrade-bundle-4.1.2.0.0.22589037.mub). In the Appliances view of all five instances, the cluster is showing as orange "unavailable". But the manager cluster is fine. A "get cluster status" shows the cluster as Stable and all manager nodes participating and "Up". The cluster VIP is working fine as well. I rebooted each manager node, but still no change. It's just the UI that is reporting it as "unavailable". I also used different web browsers. is this a bug in 4.1.2?
Replying to my own with solution: VMware Support says this is indeed a bug, and is "addressed in an upcoming version".
This fixed all 5 of my instances Why? Still scratching my head on it. (I didn't make the suggested clones. Call me crazy.)...
Workaround steps :
A. Take a backup of nsx manager (My suggestion is to have powered off clones of each manager).
B. Log in to manager node as root and take a backup of cbm jar files as below (MUST take a backup of jars before any operation)
a. mkdir jarFileBackup
b. cp /opt/vmware/cbm/cbm-app/libweb-server-cbm.jar jarFileBackup/
c. Ensure that libweb-server-cbm.jar is copied into jarFileBackup folder
C. Copy the same jar files to an additional folder to update the jar data with header size limits
a. mkdir jarFileUpdate
b. cp /opt/vmware/cbm/cbm-app/libweb-server-cbm.jar jarFileUpdate/
c. cd jarFileUpdate
d. unzip libweb-server-cbm.jar
e. cat application.properties and check the contents of application.properties
f. Append "server.max-http-header-size=32KB" to application.properties file :
echo "server.max-http-header-size=32KB" >> application.properties
Ensure that the line gets appended to application.properties
g. rm libweb-server-cbm.jar (Removing the old jar without header size info)
h. zip -r libweb-server-cbm.jar *
i. cp libweb-server-cbm.jar /opt/vmware/cbm/cbm-app/
D. Restart cluster boot manager to use the updated jar with increased header limit using below command
/etc/init.d/nsx-cluster-boot-manager restart
E. Ensure that cluster status shows correct values in UI.
Replying to my own with solution: VMware Support says this is indeed a bug, and is "addressed in an upcoming version".
This fixed all 5 of my instances Why? Still scratching my head on it. (I didn't make the suggested clones. Call me crazy.)...
Workaround steps :
A. Take a backup of nsx manager (My suggestion is to have powered off clones of each manager).
B. Log in to manager node as root and take a backup of cbm jar files as below (MUST take a backup of jars before any operation)
a. mkdir jarFileBackup
b. cp /opt/vmware/cbm/cbm-app/libweb-server-cbm.jar jarFileBackup/
c. Ensure that libweb-server-cbm.jar is copied into jarFileBackup folder
C. Copy the same jar files to an additional folder to update the jar data with header size limits
a. mkdir jarFileUpdate
b. cp /opt/vmware/cbm/cbm-app/libweb-server-cbm.jar jarFileUpdate/
c. cd jarFileUpdate
d. unzip libweb-server-cbm.jar
e. cat application.properties and check the contents of application.properties
f. Append "server.max-http-header-size=32KB" to application.properties file :
echo "server.max-http-header-size=32KB" >> application.properties
Ensure that the line gets appended to application.properties
g. rm libweb-server-cbm.jar (Removing the old jar without header size info)
h. zip -r libweb-server-cbm.jar *
i. cp libweb-server-cbm.jar /opt/vmware/cbm/cbm-app/
D. Restart cluster boot manager to use the updated jar with increased header limit using below command
/etc/init.d/nsx-cluster-boot-manager restart
E. Ensure that cluster status shows correct values in UI.
Is it related/same as this KB --> https://kb.vmware.com/s/article/97351