In iCluster main menu, the option 8. Work with the Full Cluster Status Monitor (iCluster command: WRKCSMON) is a very useful feature to help you understand the replication status. On an active node, you can use it to display the Full Cluster Status Monitor to view replication status, latency, and replication throughput of all replication groups in the cluster or for a specific named group.

 

The following are some useful pieces of information may help you when you are performing iCluster monitoring:

 

1. When you see there is high source latency but low apply latency, and the journal entries keep moving, this can be caused by:

• heavy transactions generated by application jobs
• journal scrape jobs leak of system resource on primary node to handle the transaction workload.
• communication bandwidth is not enough
• Staging store size may need to increase

2. When you see there is low source latency but high apply latency, and the journal entries keep moving, this can be caused by:

• apply job is catching up after apply suspension
• apply processing is slow due to rebuild access path

3. When you see there is high latency and journal entries are frozen, use option 5 to display the journal detail, gather the information about journal, journal receiver and sequence number, run DSPJRN to find out which transaction caused the replication to freeze or slow down:

DSPJRN JRN(JRNLIB/JRN) RCVRNG(JRNRCVLIB/JRNRCV) FROMENT(JRNSEQ)

If the journal code is not “U”, the journal type, object, or job will help you to do investigate.

In iCluster user manual Chapter 22: Monitoring, there are very detailed explanations on views, options, and status columns from "Full Cluster Status Monitor". It is recommended for reading if you are an iCluster administrator.

 

------------------------------
Linda Dan
Senior Sales Engineer, APAC
Rocket Software
------------------------------