As an app developer turned puppeteer / sysadmin,  I'm a huge advocate of 
bringing the SDLC to operations.

*Question*:  Who has accomplished this with big vendory monitoring tools - 
like App Dynamics, New Relic etc.   Anyone?

I've accomplished this while exploring Sensu. (Shameless 
plug: https://github.com/brettswift/puppet-sensu_suite/).

What does Sensu accomplish?
 - Checks are in source (even if sourced as ruby gems).  In sensu this is 
the plugin + check instantiation.   In that module it's 100% hiera, so can 
be linked to a version of infrastructure.
 - If I have a module  puppet-app1,   I can  ` include app1::sensu_checks` 
 and the app developer can supply a potential check for ops to use. 
 - dev & prod are more alike. 

Those three are key to what I want to achieve.   I don't want this  "deploy 
server and then swivel chair to add it to monitoring".    The first thing 
people will say there is "you can have agents auto-register to the App 
Dynamics console".    Or have puppet configure the New Relic agent. 

*But... * what if the topology of the application changes...  so must your 
monitoring check.  What if you have multiple versions of infrastructure and 
app code out there.......  you can't just have one check.. (think of 
something simple like a service name change, or process that now runs under 
a different user). 

... it'd be nice to have this tied to infrastructure version and come right 
from development (merged with ops checks like disk/memory etc).  

App Dynamics has a RESTful api,  so some version of the 'puppet device' 
model might work, but idempotency checks against an API could get heavy. 


We are a big MSO in Canada, with a very complex technology stack.  Hundreds 
of applications out there that all come from different departments - so the 
complexity of what we want to monitor is also high.   Hence: SDLC will 
help! 

I love my Sensu workflow.  But I may see limited traction in adoption, and 
may be forced to use something we sunk a lot of money into.  

So tell me - how did you do it?   Does anyone share my perspective?  Or do 
people resort to "just add monitoring in prod after everything is working"?

-- 
You received this message because you are subscribed to the Google Groups 
"Puppet Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to puppet-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/puppet-users/a0dad6e0-bf29-4967-b654-11339e8c10b2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to