You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Found another discrepancy between Octavia and Big-IP behavior.
When user specifies an alternate monitor IP / port for a member, Octavia treats this as just an attribute of the "main" pool-based monitor, not as a separate entity. But AS3 / Big-IP actually creates standalone, member-specific health monitor object, independent from the pool-based monitor. This member-specific monitor on the Big-IP keeps monitoring the member as long as the alternate monitor IP / port is specified for the member, even if the pool-based monitor is deleted from octavia.
The status of such member is derived from the member status on the Big-IP and it is then either ONLINE or OFFLINE, instead of the expected NO_MONITOR.
This is unexpected behavior from Openstack userbase perspective and we agreed to force remove the alternate monitor IP / port setting if the "main" pool-based monitor is deleted. This will in turn delete the standalone, member-specific monitor object form the Big-IP.
How to reproduce:
Create LB with pool and a member
Create working pool-based monitor - observe that member status is ONLINE
Specify alternate monitor IP or port for the member - observe that member status is ONLINE or OFFLINE (depending on the monitor outcome)
Delete the pool-based monitor - observe that the member status remains the same as above, instead of changing to NO_MONITOR
Remove the alternate monitor IP / port from the member - observe that the member status changes to NO_MONITOR
The text was updated successfully, but these errors were encountered:
Found another discrepancy between Octavia and Big-IP behavior.
When user specifies an alternate monitor IP / port for a member, Octavia treats this as just an attribute of the "main" pool-based monitor, not as a separate entity. But AS3 / Big-IP actually creates standalone, member-specific health monitor object, independent from the pool-based monitor. This member-specific monitor on the Big-IP keeps monitoring the member as long as the alternate monitor IP / port is specified for the member, even if the pool-based monitor is deleted from octavia.
The status of such member is derived from the member status on the Big-IP and it is then either
ONLINE
orOFFLINE
, instead of the expectedNO_MONITOR
.This is unexpected behavior from Openstack userbase perspective and we agreed to force remove the alternate monitor IP / port setting if the "main" pool-based monitor is deleted. This will in turn delete the standalone, member-specific monitor object form the Big-IP.
How to reproduce:
ONLINE
ONLINE
orOFFLINE
(depending on the monitor outcome)NO_MONITOR
NO_MONITOR
The text was updated successfully, but these errors were encountered: