Question
Capgemini
CA
Last activity: 11 Jun 2019 2:07 EDT
Agent configuration for Disaster recovery
We are preparing for simulated disaster recovery exercise and realized that handling agents add a lot of manual steps.
We have about 9 Pega servers in production. All of them are multi node setups, mostly compromising of 4 nodes each.
Now each server host multiple applications, almost all of them have couple of agents configured per application. Few are advanced which are supposed to be enable to run only one node.
While preparing DR run book. We have been told that going to each node and then starting or stopping each agent is way too much of manually work and will eat up a lot of precious time.
Wanted to see if there are any guidelines or ways to reduce manual steps involved.
Imagine having to Stop/Start about 50-100 agents on different nodes, that too in a DR kind of situation.
Appreciate any inputs in this regards.
***Edited by Moderator: Lochan to update platform capability tags***
Pegasystems Inc.
US
Hi,
The Agent Management landing page allows you to restart all of the agents in a cluster.
https://community.pega.com/sites/default/files/help_v82/pxadminstudio/procomhelpmain.htm
While on the "Cluster" tab, you can select the agents you want to restart (there is also a select all option) and click the button for the desired action (start, stop, restart, etc).
Alternatively, while I haven't personally used this, the below post has some information about stopping and starting agents programatically that may be useful.
https://collaborate.pega.com/discussion/configuring-button-agent-stop-start
Hope this helps!
-Kurt
Proximus
BE
Hello,
How do you go DR?
You start your applications on full new DR infra or you consider loss of 1 or 2 servers?
Thanks
Anthony
Proximus
BE
Hello,
on our side, to identify quickly which agent should run on one node only, we renamed them with _SN (Single Node). (_MN for Multi node, _OD for On Demand agents)
Most of our Single Node agents run on same server. If we lose it, we can start all SN on another server.
via SMA or AES, we can quickly identify/start/stop them.
if you want to spread your single node agents, you could name them SN1, SN2, etc to identify their master server. Based on the server you lose you know the ones to start somewhere else.
we've got lots of agents and this is indeed not that easy to easily manage them and even more complex to get something automated.
Regards
Anthony