How to edit my configuration for a multisite indexer cluster so that search factor and replication factor are met?

[sidekix24]

We just migrated over to a multisite indexer cluster and we're seeing "Search Factor" (SF) and "Replication Factor" (RF) are not met. Below is the server.conf setting we have on the master node:

[clustering]
cluster_label = *****
mode = master
#pass4SymmKey = *****
pass4SymmKey = *****
replication_factor = 4
available_sites = site1,site2
multisite = true
site_replication_factor = origin:2,total:4
site_search_factor = origin:2,total:4 

The setup is 2 indexers per site and there are 2 sites. As you can see HA is a big deal here so they wanted to max out the RF and SF.

We're also seeing a bunch of fixups with the message below:

Missing enough suitable candidates to create replicated copy in order to meet replication policy. Missing={ site1:2 }

I've seen some posts that say changing the replication factor to 2 fixed this issue in the past. If we do that, would leaving the site RF at "site_replication_factor = origin:2,total:4" still give us 4 copies across the 4 indexers?

Thanks

[sidekix24]

Hi,

Somesoni2 was able to provide the answer below. The second question that we have is

would leaving the site RF at "site_replication_factor = origin:2,total:4" still give us 4 copies across the 4 indexers?

Can anyone confirm this?

Thanks

[skoelpin]

Out of curiosity.. If you claim that HA is a big deal then why not add an additional indexer per site and cluster them?

After re-reading this, I'm assuming you have a multi-site index cluster?

[sidekix24]

yes...it's a multi-site cluster with 2 indexers as each site

[somesoni2 · Accepted solution]

This is a common issue that may arise when migrating from single site to multisite indexer cluster. In your each site, there are only 2 nodes and replication factor (which you set for your single site cluster) is 4. Even though the single site indexer cluster specific attribute 'replication_factor' gets overridden, it has to be set to a values that is feasible considering that site as single site cluster. So, you would need to set it to match (or lower) the number of nodes in the site, i.e. 2, and the message should go away. See this link for more details.

http://docs.splunk.com/Documentation/Splunk/6.5.1/Indexer/Bucketreplicationissues#Multisite_cluster_...

[sidekix24]

Thanks somesoni2,

Can we leave this setting the same and still get a copy on each indexer in the multisite cluster?

"site_replication_factor = origin:2,total:4"

Thanks again

[somesoni2]

Yes, since you've multi-site clustering setting in place, the site_replication_factor will take effect and each site will get 2 copies of data and with two indexers on each site, each indexer should have a copy.

[sidekix24]

Awesome...thanks