Aws-load-balancer-controller: Version 1.0 status and current image version

[jkemp101]

I'm trying to get a handle on the current status of version 1 of the controller. Things have been working fine with my limited production use but I need to decide if I want to roll it out to a broader set of services.

1. What version should we be running? The Readme still says the current version is 1.0-alpha.3, I have been running 1.0-alpha.8. I'd do a PR to update the Readme but I honestly don't know what version we should be testing to get ready for a 1.0 release. Maybe if this is still alpha we should just be running latest?

2. The Roadmap doc points to the Version 1 milestone which has 4 issues in it but none of them seem to have any activity in the last few months. Are they still the only issues remaining to have a V1.0 release? I ask because this issue https://github.com/coreos/alb-ingress-controller/issues/368 seems important and should probably be added to the milestone once it is confirmed.

3. Is the project still in Alpha? Are there near term plans to have a feature lock and get a Beta version out?

I'm just looking for some info so I can make better planning decisions. I really want to stop using CLBs.

[gaganapplatix]

We would also like to use this as well. Currently testing master which is on 1.0-alpha.9 but it is unclear when the alpha version change. Is there some release schedule or milestone around 1.0 so people can use specific tags?

[pawelprazak]

Couple of questions:

• Is there any ETA for 1.0?
• Is it actively maintained/developed?
• Is it used in production?

I'm trying to assess whether or not my team should try to adopt it or is it not actively developed.

[bigkraig]

Hey @pawelprazak, those are some pretty valid concerns.

As you can see, there has been some rough roads on the maintenance side lately. At Ticketmaster I have been focused on less Kubernetes related projects which has taken away much of my attention and @joshrosso left CoreOS and has yet not been replaced on the CoreOS/RedHat side. We did pick up another maintainer from AWS, @andrewtandoc, and he is getting up to speed and focusing on code restructure/test cases before we hit the 1.0 milestone.

Ticketmaster has been using this in production and while I don't have any official list of who else is using it, there is quite a lot of pull activity in the Quay repository.

Officially, AWS, RedHat, and Ticketmaster are committed to supporting the controller, it's just going through a bit of a transitional phase.

[ehbello]

Which version does TicketMaster use in production? Which version is currently considered the most stable?

[bigkraig]

@ehbello I just checked and the team is using 1e8a381

[pawelprazak]

Good to know! Would it be possible to release it then as a "named/tagged" release, like alpha 9?

[bigkraig]

@pawelprazak Gonna get to this soon, right now things are getting ready for the ownership transfer to sig-aws.

In the mean time the team was getting ready to change out to this build due to a regression: https://github.com/coreos/alb-ingress-controller/pull/384

[bigkraig]

403 fixes some TG node list regressions and I think will then be alpha 9 ready

[bigkraig]

I just cut 1.0-alpha.9 🎉

[ehbello]

Great work @bigkraig! I tried the lastest build f79a092 and it seems to work flawlessly. At least I don't see any errors with the debug output enabled. Thank you!

[mvillumsen]

@bigkraig Did you push the new image to the repository as well? 1.0-alpha.9 does not seem to be available at quay.io/coreos/alb-ingress-controller as mentioned in #412

[bigkraig]

@mvillumsen Sorry about that, I had updated the image url when I moved the repo over to kubernetes-sigs by accident. 1.0-alpha.9 is available now.

[bigkraig]

1.0-beta.2 has been cut and has a lot of the major rework that I have wanted to get done included. I've also gone through a lot of the miscellaneous reported bugs here and verified as many as I can that they are resolved.

I have one last feature request in #405 that I think is important to get in before a 1.0 release, since it has the possibility of changing the default routing behavior (only to one that is expected, at least)

[jkemp101]

@bigkraig Thanks for all of the changes/fixes. Looks like a lot has gotten done recently. I'm wrapped up in some other projects right now but looking forward to trying 1.0.

[bigkraig]

closer to the 1.0 release. the 1.0-beta.4 image supports the standard Kubernetes ELB tagging conventions for discovery, supports the AWS CNI which adds pods directly to the TG, and fixes some healthcheck problems people were experiencing.