Deployment Architecture

Any know issues with deployment server and master on same machine ?

Communicator

Do any one know if there are any know issues for using same Splunk instance for deployment server and clustering master ?
If this is a bad approach why ?

Contributor

I have a DS and a Cluster Master running on the same host (a VM). I created them as separate Splunk installs (/opt/splunk_clustermaster and /opt/splunk_deploymentserver). As long as they don't conflict for ports and aren't in contention for CPU/RAM, there is no reason not to co-locate them for small to medium sized deployments.

When DS gets really busy, this may become a problem. Like jtrucks says, either back down the frequency of checkins for DS clients, or setup additional DS instances on other hosts.

Contributor

The issues with port conflict only appears if you run the DS and CM as separate splunk processes. If you run them both in the same process, then there is no port conflicts.

0 Karma

Communicator

Thanks for the response. Can you be more specific regarding the port conflicts ? Currently I tested it in my development environment with a single Splunk service acting both as a deployment server and master node with out any issues.

0 Karma

Splunk Employee
Splunk Employee

Really this is about performance, so this depends highly on your indexing volume on that machine, number of hosts calling home to the DS and how often they call in, and how many concurrent searches happen on that machine. Compare all this with the hardware you have and see if it works for you.

You can tune the DS by spreading out the calling home interval, as well, from the default 30 seconds to several minutes and it can drastically reduce your performance hit as a DS.

Communicator

Thanks for the response but Master does not index any data on that machine.

0 Karma

SplunkTrust
SplunkTrust