When running the Load Balance Index Servers workflow, are there any precautions to take into consideration? Should any of the services on the main CommCell be suspended? Should this be done when there is a window where there are no jobs running? What is the best practice before running this workflow?
Solved
Load Balance Index Servers: Precautions, impact to running
+1Best answer by Marcelo Mansano
Hi Jonathan,
Thank you for your questions, it will help us enrich our documentation with missing information 
.
Answering them:
- Are there any precautions to take into consideration?
The main precautions to take here is to verify that you have other Media Agents that can receive the load from overloaded ones. We also recommend that the overloaded Media Agent to have at least 5% of total disk space of index cache disk free to be used during the migration process (copy from the old to the new server).
There is no need to disabled schedules or stop any jobs during the load balance process, the index backend will take care of locking and making sure everything is protected during the migration process.
- Should any of the services on the main CommCell be suspended?
No, that's not required or recommended in this situation, the CommCell needs to be up-and-running for the workflow to succeed.
- Should this be done when there is a window where there are no jobs running?
No, that's also not required, although it can help accelerate the process, there is not need jobs to suspended during the workflow run.
- What is the best practice before running this workflow?
There are three points we should observe before running this workflow:
1. Certify that the overloaded index server's cache disk should have ~5% of free disk space.
2. Certify that the overloaded index server has shared data-path with other Media Agents, so we can use those for index migration. This can be done by verifying the primary copy of the existing Storage Policies' data-path and check if there are more than one Media Agent there.
3. Certify that the index servers are connected and ready (Commvault processes are up-and-running), and that they can communicate with themselves.
You can always check the "Index Server Load Report" to see the status and what's the recommended management steps there.
As mentioned by wgrande, we have a default schedule that runs the workflow once every month (last Sunday), but I can recommend you to create a new or change the schedule to run the workflow weekly - this way you won’t need to worry with index cache space usage.
+23- Vaulter
- 4996 replies
Hey
I’m including some documentation here which covers the instructions though there aren’t any explicit best practices:
https://documentation.commvault.com/commvault/v11_sp20/article?p=100933_1.htm
You don’t want to suspend any Commserve services (since the operation and web server need the CS online), though there’s no need to suspend jobs or anything. The operation waits until the index is free before doing anything on its own.
As the link mentions, this normally runs at midnight on the last Sunday of the month though you can schedule it to run weekly.
Let me know if you have further questions.
+10- Vaulter
- 107 replies
Unless you’ve added a new servicing Media Agent, a new hard drive into an existing Media Agent (to store your indexes on), or added a fair amount of clients within the past 30 days… you shouldn’t need to manually run this workflow.
If you did or you want to run it, you won’t be required to shut down any services before running it against your Media Agents. I would say you should run it without backups running but it won’t harm the index relocation process or backup operations if both were performed at the same time.
+1Thanks wgrande, I took over this Commcell about a year ago and the previous person had not keep up on maintenance. I am finding little things like this as I learn more about how to keep things running well and noticed I had 3 of my media agents in a waring status. 2 VM media agents and one physical. I have been over the past year cleaning up things and getting things up to current levels. Just trying not to break anything in the process.
- Vaulter
- 2 replies
- Answer
Hi Jonathan,
Thank you for your questions, it will help us enrich our documentation with missing information .
Answering them:
- Are there any precautions to take into consideration?
The main precautions to take here is to verify that you have other Media Agents that can receive the load from overloaded ones. We also recommend that the overloaded Media Agent to have at least 5% of total disk space of index cache disk free to be used during the migration process (copy from the old to the new server).
There is no need to disabled schedules or stop any jobs during the load balance process, the index backend will take care of locking and making sure everything is protected during the migration process.
- Should any of the services on the main CommCell be suspended?
No, that's not required or recommended in this situation, the CommCell needs to be up-and-running for the workflow to succeed.
- Should this be done when there is a window where there are no jobs running?
No, that's also not required, although it can help accelerate the process, there is not need jobs to suspended during the workflow run.
- What is the best practice before running this workflow?
There are three points we should observe before running this workflow:
1. Certify that the overloaded index server's cache disk should have ~5% of free disk space.
2. Certify that the overloaded index server has shared data-path with other Media Agents, so we can use those for index migration. This can be done by verifying the primary copy of the existing Storage Policies' data-path and check if there are more than one Media Agent there.
3. Certify that the index servers are connected and ready (Commvault processes are up-and-running), and that they can communicate with themselves.
You can always check the "Index Server Load Report" to see the status and what's the recommended management steps there.
As mentioned by wgrande, we have a default schedule that runs the workflow once every month (last Sunday), but I can recommend you to create a new or change the schedule to run the workflow weekly - this way you won’t need to worry with index cache space usage.
Thanks!
+1Thank you all for the quick response on this question. I see that this index cache load balancing has been enabled for the last Sunday of each month. I was worried that it might not have every been implemented by my predecessor. I installed the report for the Index Server Load and the recommendations on these Media agents are to add space. On the VMs (media agents) this is not a issue, but with a physical Media agent this poses a more complicated issue. The physical media agent is on our main media agent and acts as the primary media agent for the entire CommCell with a secondary at another physical location. Both Aux copy between each other for DR purposes.
1 person likes this
+23- Vaulter
- 4996 replies
Glad to see everyone was able to help 
Reply
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.