List of System Nodes Outdated in Designer Studio and AES
We frequently replace our Pega JBoss Application server nodes. This causes a new node to be created in the System Nodes list. However, the old nodes still show up with the node id's that were previously existing. Is there a way to cleanup these stale instances of Nodes.
Go to Designer Studio--->System--->General--->Systems,Nodes,Requestors
There is a Nodes list at the bottom
Accepted Solution
Amazon Corp
US
There is a OOTB activity 'RemoveOldNodes' that you can run to remove the old nodes from the list.
Pegasystems Inc.
IN
Hi Abhishek,
You can tally these Node IDs in table pr_sys_statusnodes .
You can review Data in this table and take appropriate decision for maitainability.
Amazon Corp
US
Thanks for the response Arvind. I am able to see them in the table mentioned, but they also have records for old nodes which do not even exist. The list in Designer Studio and this table tallies, but how do I cleanup old nodes? Is it safe to just drop those records from this table? Or is this table just a list but the actual information comes from somewhere else?
Pegasystems Inc.
IN
I assume, this is development environment, so you can play around safely.
you can take backup of data of this table, delete the content of table and restart the server.
Re-starting of server makes current node entry in this table.
kindly post the results after performing this test.
Pegasystems
IN
The reason to keep the node information in AES is to be able to know which node generate the alerts or exceptions. Deleting the node details will not give you that detail. That said, based on your use case, you can delete the specific nodes which are no longer in use. Pega platform has no OOTB mechanism because we don't know if the node has been shut down and will be restarted or the node will no longer ever come back up.
Amazon Corp
US
Yes it makes sense that Pega may not know if the node has been shut down and will be restarted or the node will no longer ever come back up. But I am sure the nodes list that shows up should be coming from either in some file on Filesystem or some Table, so I was looking to find out what that location is so that I can bulk manage it.
How does creating a new domain name (keeping the old one as is) affects the System Nodes. In our case new System Nodes got created and when the app server is bounced, one of the existing NodeIDs is picked up randomly. Why is it picking a different NodeID every time we bounce the app server? This keeps the search functionality breaking (unless we add multiple Host Node IDs under 'Search Index Host Node Setting). Just for a single node environment, I see 4 System Node instances (big mess) and I see proportionate number of Agent Schedule instances for all agents with an X factor of 4. I see two records in pr_sys_statusnodes table. Appreciate if someone could share ideas on deleting these System Nodes, What gets affected if we get rid off the old domain name and does it have affect on WebSphere profiles etc?
Pegasystems Inc.
US
Recently needed to truncate pr_sys_statusnodes on 7.2.1 to resolve an issue with JBoss startup ("This node not found in the database - Either the record was never saved or was deleted. Unable to join the cluster "). This resolved the issue. However, after logging in and viewing the "Systems, Nodes, Requestors" landing page, we see the old node is still referenced. This article references using RemoveOldNodes activity to clean up this data. But, this activity appears to delete records from pr_sys_statusnodes - which had been truncated.
Recently needed to truncate pr_sys_statusnodes on 7.2.1 to resolve an issue with JBoss startup ("This node not found in the database - Either the record was never saved or was deleted. Unable to join the cluster "). This resolved the issue. However, after logging in and viewing the "Systems, Nodes, Requestors" landing page, we see the old node is still referenced. This article references using RemoveOldNodes activity to clean up this data. But, this activity appears to delete records from pr_sys_statusnodes - which had been truncated. Furthermore, tracing this landing page, the data is coming from Data-Admin-Nodes which maps to pegadata.pr_data_admin. Is there another activity to clear this table?