Troubleshooting teamed environments
Controllers dropped from team or unable to form team
Symptom
A group of SDN controllers fail to form a team, or one or more controllers are dropped from the
configured team.
Cause
The system clock on one or more controllers are not synchronized with each other.
Action
1.
You must synchronize all systems (even after a power cycle).
2.
If the time is incorrect on boot up, restart the
sdnc
/
sdna
service.
3.
If you are installing the controller on a Virtual Machine (VM), you must synchronize the host
hypervisor with the NTP and NTP client running on the hypervisor host.
4.
Because there can be differences in the time reported by different NTP servers, Hewlett
Packard Enterprise recommends that servers in a team be configured to use a centralized
NTP daemon. If the servers for the controllers in the team are configured such that they
connect to different NTP servers, change the configurations of the servers to use a centralized
NTP daemon.
Teaming framework does not run
Symptom
The teaming subsystem does not run.
Cause
The
Iptable
rule programming for the teaming framework (Hazelcast) failed.
Action
For each controller in the team, do as follows:
1.
Enter:
sudo iptables -D OUTPUT -p tcp --destination-port 5700 -j DROP
2.
Enter:
sudo iptables -D INPUT -p tcp --destination-port 5700 -j DROP
3.
Restart the controller.
Controller becomes suspended
Symptom
The controller is suspended and the RSdoc and UI are unavailable.
Cause
A controller transitions to the
suspended
state when quorum is lost or core services report critical
health. When a team is partially created, a controller might never get quorum, and thus remain
suspended, because team creation failed in other controllers.
Troubleshooting teamed environments
163