Raft State Machine

[verbal_666]

Hi.
I do not understand well the SHC config,

[raft_statemachine]

disabled = <boolean>
* Set to true to disable the raft statemachine.
* This feature requires search head clustering to be enabled.
* Any consensus replication among search heads uses this feature.
* Default: true

replicate_search_peers = <boolean>
* Add/remove search-server request is applied on all members
  of a search head cluster, when this value to set to true.
* Requires a healthy search head cluster with a captain.

 What changes in a SHC by setting "disabled = true or false"? By default is true.

"replicate_search_peers = true" works only if disabled is false.

What does setting this to true or false do to the cluster?

[PickleRick]

https://en.m.wikipedia.org/wiki/Raft_(algorithm)

Without raft algorithm your captain election will not work properly. You might get away with static captain but it's not fault tolerant so if you lose your static captain your SHC will more or less fall apart.

[verbal_666]

I neved had problems with Captain Election, both with

[raft_statemachine]
disabled = true
disabled = false

🤷‍

[PickleRick]

It's not the only factor in captain election. So just because you have raft enabled doesn't mean that your election will work properly.

[sainag_splunk]

@verbal_666 May I know the reason you are tryin to update this?

Basically replicate search peers is when you want to  add a non-clustered indexer via the GUI, replicate that to your other SHC peers, I don't think that required. 

Refer: https://docs.splunk.com/Documentation/Splunk/9.2.0/DistSearch/Connectclustersearchheadstosearchpeers...

I have only seen  replicate_search_peers is used where you have search head cluster is in use with no index clustering, so if you want to search the non clustered indexers, its set to false.  If you want it to replicate its set to true.

replicate_search_peers = true

• Only works if disabled = false
• Automatically copies search peer connections to all SHC members
• Most useful when connecting SHC to non-clustered indexers

If this helps., Please Upvote

[verbal_666]

Don't understand the question 🙄🙄🙄

I have a non clustered Indexing infrastructure (8 standalone indexers).

---

@sainag_splunk wrote:

I don't think that required. 

I decide what's required 😉and i want the SHC to replicate distsearch.conf 😉😉

My question was another, leave replicate_search_peers away,

---

What changes in a SHC by setting "disabled = true or false"? By default is true.

---

[sainag_splunk · Accepted solution]

The disabled setting in SHC only impacts captain election and member roster management.

[verbal_666]

---

@sainag_splunk wrote:

The disabled setting in SHC only impacts captain election and member roster management.

---

Ok, so it's minimal, and has ne real impact of cluster operativity 👍

Thanks 👍