Question: Heyo, autoscaling q: if I update a launch configuration, is there a way to automatically trigger my ASG to replace the instance using my new launch configuration, or do I just kill the instance manually and let the ASG fire up a new one? Looks like I forgot the cidr blocks section in my egress rules... Bloody security groups balls, my egress rule in the SG isn't showing up hm, did it manually, but now my target group seems to be hitting port 80 instead of the assigned ephemeral port of my container in ECS

Asked By
phroggyy
Asked At
2018-03-10 15:42:07

Found 15 possible answers.

User Answered At Possible Answer
rlister 2018-03-10 18:31:07 @phroggyy If you just update your cloudformation stack with any launch config change, the ASG will update. If you have an appropriate UpdatePolicy on the ASG, it will do a rolling replace of instances.
phroggyy 2018-03-10 19:16:42 Oooh @rlister what would that update policy look like? Haven't delved into that yet but feel like I should add that
rlister 2018-03-10 19:20:08 @phroggyy https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-attribute-updatepolicy.html
phroggyy 2018-03-10 19:58:40 Thanks mate
AWS_News_Blog 2018-03-11 17:40:20 AWS Summit Season is Almost Here – Get Ready to Register! http://feedproxy.google.com/~r/AmazonWebServicesBlog/~3/-lcJZH2g-aY/ I’m writing this post from my hotel room in Tokyo while doing my best to fight jet lag! I’m here to speak at JAWS Days and Startup Day, and to meet with some local customers. I do want to remind you that the AWS Global Summit series is just about to start! With events planned […]
hello390 2018-03-11 19:42:08 Hi guys, Can someone with more experience tell me if that’s still the case in 2018? Gotta decide now between serverless and EC2. I’ve just read this article https://www.datawire.io/3-reasons-aws-lambda-not-ready-prime-time/ and I was wondering what’s your opinion on AWS Lambda? The author is rather skeptical and says there are many issues: poor error handling, a lot of boilerplate code.
jordan 2018-03-11 19:44:48 uh, I think 2 years later most of these arguments aren't the case, e.g. bad documentation and logs not making it to CloudTrail And if you couple Lambda with Step Functions, you have an even more powerful set of tools API Gateway has much tighter integrations with Lambda now, too
hello390 2018-03-11 19:51:31 Cheers @jordan so Lambda won’t be troublesome for an agile startup.
jordan 2018-03-11 19:57:18 If you're looking for an API, nope
phroggyy 2018-03-11 22:25:49 Only had time to look at this now @rlister – do you know if this is possible to setup using terraform? This seems to suggest it's not https://github.com/hashicorp/terraform/issues/1552
rlister 2018-03-11 22:29:12 @phroggyy Not used terraform in quite a while, so I can’t say for sure. But seeing as the rolling updates are handled by the cloudformation service, I suppose this is not surprising.
phroggyy 2018-03-11 22:32:09 ah, didn't realise that's handled by the CF service itself running a quick ls -la /var/log/application on the instance tells me the log file (which I create in user_data on launch) is owned by root, and if I try to chown www-data:www-data /var/log/application/app.log , it fails since that user doesn't exist I have an ECS service running a webserver with the www-data user. I want this user to be able to write to the *container instance* file /var/log/application/app.log . Now, it's easy to just mount a volume in my container defs for this, basically just mounting the /var/log/application directory to /var/www/html/logs (where the application writes its logs to). However, I can't seem to figure out how to give the *container user* ( www-data ) permissions to write to the host Having a bit of an issue here with ECS, files and permissions. Scenario: Yeah makes sense then
rlister 2018-03-11 22:37:20 You could roll your own with lambda, but might hit 5 minute limit if it takes a while to time out healthcheck failures. Bear in mind you need to hang around long enough to handle the rollback case.
phroggyy 2018-03-11 22:39:00 hm, right :confused: I mean I can always just let my deployment script deal with it manually, just run it on deploy write in the pipeline was just hoping for an easier solution
rlister 2018-03-11 22:42:52 Maybe ask in #infra-as-code , see what other terraform users do for this?

Related Questions