What is Maya ChatOps?
Maya-ChatOps is one of the core areas of MayaOnline, covering the storage operational support of kubernetes clusters. DevOps developers and admins get the alerts and analytics of their OpenEBS volumes deployed across multi-cloud kubernetes clusters right into their slack channels. Our vision of ChatOps extend beyond just simply providing alerts and providing a way to query any configuration and status from slack. It goes all the way to interact with DevOps developers and admins to manage the yaml config files in their CI/CD system.
What is MuleBot ?
MuleBot is the name of the bot or slack application from Maya. MuleBot is a distributed slack application. MuleBot responds to user’s queries about configuration and status of the OpenEBS volumes. Sometimes, MuleBot tries to surprise you with smart alerts prior to a real situation happens.
How to use Maya ChatOps?
You can start using ChatOps by adding Slack integration in MayaOnline. The MuleBot slack application will be installed in your workspace. Subsequent steps involve configuration a single or multiple clusters to the desired slack channel. This mapping is maintained in the MO in the form of a “slack-card”.
You can add as many slack cards as you want for your clusters. Through this channel you will be able to interact with clusters imported in the MayaOnline.
What are we using underneath for powering our ChatOps?
Well, this is why I am writing this blog, to tell you the various choices we had and why we ended up with a particular choice. Some of the design goals we kept while choosing the bot framework are:
- Users of MayaOnline will be in thousands to begin with, so, the bot framework has to be multi-tenant
- The bot has to be a micro service and suitable to run seamlessly on kubernetes
- The bot framework has to have the NLP AI support for us to get that capability out to the users in the near future
So, we looked at Hubot, StackStorm/errBot and BotMan
Each one of them had their benefits but none of them were multi-tenant. Then we looked at which is easiest to add the multi-tenant support to, BotMan came surprisingly easy to add this support to. BotMan is thin, and is written as a stateless application. The preliminary approach involved passing the user configuration in environment variables. All it needed was a thin shim to get user-config details dynamically and we had achieved multi-tenancy ! It is that simple.
We kept a combination of slack team_id, channel-id as the key of the mulebot to manage the link between the slack user config and MayaOnline user config.
With the above design Maya ChatOps allows users to configure different slack channels for different kuberenetes clusters at MayaOline.
Next steps:
- In the coming weeks, we will try to post some example scenarios suggesting how smart our MuleBot can be 😄
- We plan to extend the BotMan framework to provide Maya ChatOps API. A DevOps developer/admin can use these APIs to integrate better into their CI/CD
- Currently it is integrated with Slack. PagerDuty is on our horizon.
