Fork me on GitHub
#clojure-uk
<
2017-11-10
>
dominicm07:11:50

@mccraigmccraig curious to know how you've found cloudformation?

dominicm07:11:04

I'd also like to know (This wasn't clear from documentation) where does the cf template run? Like, if I tell it to start 15 servers, wait 1 minute, then turn off my laptop. What happens?

thomas08:11:33

moin moin morning

mccraigmccraig08:11:26

@dominicm i used mesosphere’s cf template generators, so my exposure to the minutæ of cf templates has been minimal - i just deploy a stack from the generated templates

mccraigmccraig08:11:28

the templates run in the cloud, with detailed event reporting for each step. it can still be pretty opaque when things go wrong though

mccraigmccraig08:11:13

i chose this way rather than some terraform because it gives me an out-of-the-box dc/os cluster which supports the stuff which will make future life easier (particularly scaling groups) and is supported by mesosphere

mccraigmccraig08:11:02

there are some limitations built into these CF templates though, e.g. masters are deployed from the same AMI as agents, which means my masters have a load of unused storage volumes attached, but i can live with that

dominicm08:11:55

@mccraigmccraig I was surprised that most of the errors couldn't be caught aot by aws

mccraigmccraig08:11:44

they probably can... i had one which took me ages to figure out, but was caused by me not reading the ample warnings about giving an iam role with sufficient privileges. PEBKAC

dominicm08:11:44

Where were the warnings? AOT or nested in some documentation?

mccraigmccraig09:11:23

during the cloudformation stack deploy screens... it was all new and unfamiliar and i glossed over them completely, although they are spectacularly obvious with hindsight

danm09:11:46

@dominicm Once the template has been pushed to AWS, all the work is on AWS

danm09:11:06

So you can happily turn your machine off and it'll still spin up the instances defined by the template etc

danm09:11:42

Generally when libs and such are waiting there they're periodically polling AWS to see if the create was successful or not, because it can fail and the template will just sit on AWS in FAILED state

dominicm09:11:03

@carr0t perfect! I wanted to have a machine with its own login system, that had complex authorisation policies on cf templates.

danm09:11:10

Like if you asked it to spin up more resources than you have available or whatever

danm09:11:26

Or if you ask it to do something that it turns out you don't have the privs for

yogidevbear11:11:19

How is everyone doing today?

yogidevbear11:11:58

Any of you have the fortune of an open source Friday policy at work?

danm11:11:10

Other teams certainly have a 10% day once per sprint, but we have too much on for clients to do that 😞

yogidevbear11:11:48

I feel you pain 👍 Would love to convince the powers that be to give us a few hours a week to do something like that

danm11:11:16

Because we have so many more clients and bigger systems than most teams here what would otherwise be 10% time is spent ensuring our docs are still good and usable and there, both client facing stuff and internal process stuff

otfrom16:11:57

I like to do open source every day if I can.

yogidevbear17:11:50

But you pay the salaries @otfrom;) You're allowed to during work hours 😉

otfrom17:11:31

@yogidevbear everybody else does too (tbh, they do more than I do, I'm really mostly post-technical now, I need to find a management slack)

seancorfield18:11:57

I find I work on open source a lot more when my job involves less coding (management, pure architecture). If my job has plenty of coding opportunities, my open source work suffers!