docker-registry-proxy
docker-registry-proxy copied to clipboard
Add DISABLE_IPV6 to ignore AAAA records when proxying requests
I was getting signal 11 crashes from nginx with
docker-registry-proxy versions 0.6.2 and 0.6.4 (test performed pulling ubuntu through docker-registry-proxy without it being in the cache):
2022/10/19 03:54:57 [alert] 74#74: worker process 76 exited on signal 11
Leading to image pull errors in Docker and Kubernetes, e.g.:
docker: error pulling image configuration: download failed after attempts=6: EOF.
With the latest code in master, I get failures like instead of signal 11, but it sometimes seems to work, however:
2022/10/19 04:00:39 [crit] 83#83: *23 connect() to [2606:4700::6812:7c19]:443 failed (99: Address not available) while connecting to upstream, client: 127.0.0.1, server: proxy_caching_, request: "GET /v2/library/ubuntu/blobs/sha256:d63f752103bb93d846e17fa9996d3e708717c51b106382fe84d8527ee47a3547 HTTP/1.1", upstream: "https://[2606:4700::6812:7c19]:443/registry-v2/docker/registry/v2/blobs/sha256/d6/d63f752103bb93d846e17fa9996d3e708717c51b106382fe84d8527ee47a3547/data?verify=1666155039-2OQ%2BkxWXW53KIWSACLv3vpuPCD8%3D", host: "registry-1.docker.io"
If I set DISABLE_IPV6=true with this change, it works just fine, no errors like shown above.
Test workflow failed also against the master branch. Something possibly unrelated broke...
Test workflow failed also against the master branch. Something possibly unrelated broke...
Doh! Sorry, I missed your comment. I'll take a look shortly.
Hi @deejgregor ! Did you have a chance to take a look at the check?
Hi, any chance we can get this fix released? Thank you.