Further to Q2, I can imagine how to configure two different EC2 targets using the config file to achieve Blue/Green pretty easily. What I’m raising is that I initially thought that this kind of deployment process was included in Roll because of the tagline. Is that my mistake or will Roll eventually have a sample of full Blue/Green to support the tagline?
Hi @steveb8n
good morning…
The way we do blue green is to have multiple asgs running diff versions
Then you flip which one has the load balancer when you are happy
I figured it would be something like that. I’ll try setting that up
would be good to have that particular recipe in the docs for all
Totally :-)
The idea is you keep modifying the config to promote different versions, so you have a deployment history
likewise the KMS secret read sample
interesting. I discovered the use of tags in your sample to generate artifact names in the S3 bucket
would that not work i.e. then wouldn’t need config changes
?
This feedback is great and we will work on the docs
Tags? Do you mean :latest ?
I found that the upload
command needed a git tag to work. when I added a tag, everything worked
so I inferred that git tags were the way you ‘version’ your artifacts. Is that inference wrong?
in my case I use semvers as tags
Ah I see. Yes, it shouldn't need tags. I http://am.in the office today so will have a look
good to know. want me to log an issue for that?
It should be able to use the git sha and not rely on tags
Please yes!
done
Cheers
Now that I am fully deployed, I’m wondering how else I can use Roll. e.g. Datomic. Or should I use the packer recipe for that?
I will write you something up today/tomorrow
also, should I log an issue for Q1 as well?
For Q1, try doing a tf plan in between
ok, will do. I think I already have but I’ll double check
before long I’ll build a docker image so I can run all this in Bitbucket pipelines. that can be public so all can benefit
It might be terraform has lost track of an asg. I don't think this is a roll issue..
That would be great
last Q (I promise). I also want to deploy a Reimann server and hook up clients. Do you think Roll will be a good fit for that as well?
The approach of add tf config in addition to what roll has built.. Yes. We do this for Riemann
Let me get something to you
great. happy to test all this great stuff. thanks again, this is fun
We have a pack-riemann on github I think
Will check the status later
oh, so Roll drives the packer process? I didn’t make the connectin
connection
No it doesn't, but this repo gives an idea
cool. I’ll do the homework. thx
Have to go to the gym now... but in the office a little later
Go big or go home!
Right ;-)
It has been such a mental couple of weeks, and holidays before that. Normally I am more responsive, so apologies for that.
no problem, if I’m ever in a rush I’ll say so, but that’s almost never. just grateful 🙂
Thanks to @dominicm for stepping in. The good news is Roll is on the roadmap for a couple of clients, so it will get some more upgrades and general love
and @dominicm has been great
:simple_smile:
Ttyl
see ya.
btw I’m using Pedestal and lein to build instead of Yada/Boot so nice to know another variation works with Roll. And port 8080 instead of 3080
now I’ll stop 🙂
The AMI is now optional with roll :) and it will get the right one for your region.