We have been using Collabora Online Development Edition docker image along with Nextcloud and users have been reporting back that there is an annoying 20 second delay every time a document is opened. by running tcpdump on the docker0 interface we can see that the docker image is trying to make DNS lookup requests to 192.168.65.1 (in the first docker image release) and 192.168.100.1 in the most current docker image). upon investigation we found these values to be hard coded into /opt/lool/systemplate/etc/resolv.conf docker automatically updates the /etc/resolv.conf (by way of a mount into the container) but has no knowledge of this other resolv.conf which from what I understand is some sort of chroot style environment for running documents. we have a workaround which has been discussed here: https://help.nextcloud.com/t/collabora-connecting-delay-for-20-seconds/1881/53 essentially what we have done is to symbolically link /opt/lool/systemplate/etc/resolv.conf to /etc/resolv.conf and things work well. new users are finding this a challenge to deal with as they don't have the ability to modify docker images easily. it also detracts from the whole "docker as a container" solution anyhow once post-remediation is required. could someone please address this issue, either by implementing the solution we have come up with, or by integrating something similar so the hardcoded values in /opt/lool/systemplate/etc/resolv.conf are not being used. cheers, wizdude.
Thanks for the hint, I pushed a new image to docker hub today, which contains this fix.