Unable To Invoke Log Reader From Application Running On The Standby Manager Node (Doc ID 1528371.1)

Last updated on MARCH 03, 2019

Applies to:

Goal

Qn1:In our system the active manager logs notifications to a directory on its local hard drive. The standby manager does not have access to this
location. This means that after a manager failover the new active manager will not have access to the notifications logged by the former
active manager, correct?
 
Qn2:Does the Log reader reads records directly from a persisted log file and not from a memory file?
 
Qn3:At this point customer don't have a network store accessible by both manager nodes. If a standby manager has a local copy of the persisted log
files, can notifications be read from the standby manager node?
 
Qn4:Customer application that invokes the Log reader is a part of the application suite that constitutes the active service for the manager node
component. In the case when the component residing on the standby manager node gets an active assignment is there a way to force a cluster/NTF
manager  failover so that the active manager and the active management component are always operating on the same node?
 
Qn5:Is there a way for the application to trigger a cluster manager failover. I probably can do this by administratively locking the active manger
node, correct?
 
Qn6:The easiest way to solve customer problem would be to ensure that the active management component is always collocated with the active
Cluster/NTF manager. Is there a way we can achieve this?
 
Qn7:What's the easiest way to determine which manager node is the active cluster manager?
 
Qn8:considering using the SU ranking system, customer cannot guarantee that one management node will always be started first. After chassis reset
there is a certain variability in bootup timing of blades and even though we do give a 15-second timing preference to one ('primary') management node
over the other we still see enough cases when the 'non-primary' node comes up first.

Customer do use SU ranking for service groups other than the System Manager and we ensure that active assignments go to the 'primary' blades in each
group by imposing a significant cluster startup timeout so that all nodes in the group are up by the time AMF manager starts assigning roles.
Unfortunately, this will may not help in any way to ensure that Cluster manager operates on a particular node.?

 

Solution

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.