I recently had to decommission a couple of live services which meant disabling certain aspects before the production region was decomissioned by others. This however left pre-production still running in two regions which equated to about 2*2*35=140 services that I had to shutdown. Using the openshift UI is too slow and clunky so I opted for the right way which is to use the oc cli.
When one needs to test the connection from a kubernetes pod to an external service it can be difficult if the pod is based on an image with the bare minimal of networking utilities such as ping or traceroute. Happily enough Curl seems to be quite often available and while testing http is easy enough what do you do for other protocols?
If you use Bunyan and decide to use a MongoDB as a logging appender (destination) then you will arguably know what a great use of MongoDB it can be. Querying is easy but sometimes visibly parsing a raw Bunyan log entry JSON is a pain but luckily Bunyan CLI will parse and display these in a much easier format. However just extracting the raw Mongo documents using Robomongo and piping to bunyan will not work as the document schema and extra added Robomongo cruft differs a little making it non-parseable by Bunyan therefore we need to transform the entries a little which we can do using standard Mac or linux utils.
In a previous post I showed how to drain a queue but during some work I needed to publish some test messages so I added a new mode to publish as well. In rabbitmq you don’t publish to a queue though but to an exchange which will route the message to the appropriate queue. The new publish mode has been added to the drainer which is a bit odd however as well as accepting a cli option to switch mode a new cli script qfiller has been added to compliment qdrainer which is added to the path as part in npm -g install.