asgard
asgard copied to clipboard
Netflix
Adding the T2 class of burstable instances
Looks like line length coding-style test is the only thing here. Would really like to get this merged. Is there anything other to do?
Hi @lightcapda and @marco-hoyer ... First of all, thanks for taking the time to put this together. We're in the process of a major architectural revamping for our Continuous Delivery/Deployment tooling... Some background here: https://github.com/Netflix/asgard/issues/486#issuecomment-43157790 ... In the meanwhile, we don't have any resources dedicated to enhancing Asgard, only providing bug fixes.
That said, if you guys wanted to take the time to fix the tests and you can find somebody from the community to give some thorough testing to this feature, we can roll it up into this repo. Maybe @pas256 or @e0d would have some free bandwidth to test this branch when the build passes?
We needed to make this change to utilize t2's for our own deployments. I updated our lower environment Asgard instance yesterday and will be conducting a few tests myself. If anything odd comes up I'll delve into it.
@danveloper I can look at this a little over the next couple days, but say that it looks good, what would be the process for merging this and cutting a new release? Also, since this is a breaking change for users of legacy micros, there would need to be change log updates, etc.
@e0d you rock! Just curious, what is the breaking change here? Looks like t1 support is retained, no?
@lightcapda An issue I see here is that attempts to launch clusters using t2.micros that result in errors, either because the instance isn't in a vpc or the wrong virtualization type is in use, silently fail.
@e0d I'll take a look when I get a chance. I've mostly been launching t2's within a VPC.
I recently tested the changes in this pull-request and couldn't find any issues. Started multiple t2.* instances inside vpc without any problem. If I try to do it without vpc, there is a nice error message in the ASG events view reporting that the instance needs a vpc. Works like I would expect it to do. @e0d, can you give a deeper explanation of your problem?