You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
If you are interested in working on this issue or have submitted a pull request, please leave a comment
Tell us about your request
While upgrading to Docker 20.10.6, our team noticed that the port mappings against deployed tasks were no longer behaving the way they had been in 20.10.5.
Prior to upgrading Docker, tasks that had 1 port defined would map to the IPv4 and IPv6 interfaces on the same port. This allowed the targets, registered at the target group layer by ECS, to flow traffic properly into the containers.
After the update to 20.10.6, Docker fixed a broken IPv6 configuration in their engine, which now results in 2 random ephemeral port numbers to be applied to Docker containers. This causes Target groups to have split ports and ultimately, if IPv6 is not being supported, half of the traffic to the targets fails out.
Which service(s) is this request for?
ECS
Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?
This would help ensure that port mappings defined in task definitions are accurately projected onto the resulting container that is launched, while using the 20.10.6 docker engine.
Are you currently working around this issue?
Downgrading to Docker 20.10.5
Community Note
Tell us about your request
While upgrading to Docker 20.10.6, our team noticed that the port mappings against deployed tasks were no longer behaving the way they had been in 20.10.5.
Prior to upgrading Docker, tasks that had 1 port defined would map to the IPv4 and IPv6 interfaces on the same port. This allowed the targets, registered at the target group layer by ECS, to flow traffic properly into the containers.
After the update to 20.10.6, Docker fixed a broken IPv6 configuration in their engine, which now results in 2 random ephemeral port numbers to be applied to Docker containers. This causes Target groups to have split ports and ultimately, if IPv6 is not being supported, half of the traffic to the targets fails out.
Which service(s) is this request for?
ECS
Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?
This would help ensure that port mappings defined in task definitions are accurately projected onto the resulting container that is launched, while using the 20.10.6 docker engine.
Are you currently working around this issue?
Downgrading to Docker 20.10.5
Additional context
Attachments
moby/moby#42313
The text was updated successfully, but these errors were encountered: