I recently tried upgrading dagger and noticed my pipeline stopped working. Specifically the containers that I launch no longer can network between each other.
I've tested that the last version that is still working is 0.18.16.
I launch 2-3 containers and I manually assign them DNS names by writing out each of their /etc/hosts files. I basically have a process that looks at the initial /etc/hosts file to discover the containers assigned IP address, and then update each /etc/hosts file so that all containers know the DNS of each other.
This initial discovery process still works fine, but it appears that networking between the containers is just broken.
On 0.18.17 I see error messages like this:
lookup ujfvap1713d22.cen23482dp2bg.dagger.local on 10.87.0.1:53: no such host
lookup ujfvap1713d22.ave6bb9s0j7m4.cen23482dp2bg.dagger.local on 10.87.0.1:53: no such host
https://dagger.cloud/conor/traces/1d3df6d4a3fdca58448fd2607cd6b4c1?span=349e0d678a857952
When trying on 0.19.6, it seems it gets farther & DNS works, but the networking just fails.
dial tcp 10.87.0.80:26656: i/o timeout
(sorry this one is pretty noisy)
https://dagger.cloud/cordialsys/traces/7dc112f62f3a277410c24b21de5c52a9?span=cfb7f2344f703c27#cfb7f2344f703c27:L144
Is there a better way that I can ensure networking between containers, or can test this issue?