For docker nodes
Usually, we ask docker node operators to not update to a new version too early because of the risks. This time we need as many nodes as possible updating early to 5f2777af9-v1.50.4-go1.17.5 The reason for this is that the new docker image contains the Linux updater. We haven’t been able to test it on all operating systems and architectures. So this is the last chance to find and fix potential bugs in time before the docker image gets pushed to all docker nodes.
That is correct. The new docker image should follow the rollout cursor. At the moment the rollout cursor is telling 10% of all nodes to update. The remaining 90% will stay at v1.49.5. The next rollout step would be 25%.
Wouldn’t it be better to use an image like :beta or something that can be held up to date. Otherwise nodes may get stuck on an old image that will node be updated whenever the next update happens. (I realize that might be a while as the intention is to not update the container anymore, but you never know)
No, if it doesn’t work, report back and revert to the latest image. The node version hasn’t actually changed yet, so it’s not even a downgrade. Just a different way of getting the node software.
This suggest that the DNS is not working, as it should. Could you please add to your docker run command the option --dns 8.8.8.8 before the --name storagenode option?