I was able to confirm with infrastructure in #asfinfra slack that these blocks are in place to make the RAO server handle the load. They commented that for at least 2 more weeks, the blocks can't be removed.
Since we still have repo.grails.org, I inquired about the policy of using it for snapshots. They suggested we reach out to the VP of infra or legal. Does anyone have experience here that could help us with this request? On 2025/06/28 13:36:24 James Daugherty wrote: > As part of the forge merge PR, I've been investigating the timeouts that > have plagued our builds in Github. First, I must state that all of our > builds have been changed to only query RAO for specific namespaces and > coordinates so I do not believe we're violating any 404 policy. It's > possible I've missed something, but based on the ordering we shouldn't be > querying RAO unless there's a package to pull. Can people please check our > gradle files and confirm I haven't missed any configuration? > > With that said, based on my recent debugging I think our timeouts are due > to blocks. I think there's a large number of Github action runner IPs that > are blocked by ASF infra. > > Check this out: > https://github.com/apache/grails-core/actions/runs/15944575485/job/44977011587 > > > I changed this job to output the IP of the github action > runner: 68.154.30.153 > > According to https://infra.apache.org/abc/ it's blocked: > > BL018 <https://infra.apache.org/abc/#BL018>: Too many 404s on > repository.a.o (bad CI configs looking for maven repos) (1023 >= 1000) > > Because we're using github actions, we're using a shared platform. It > seems that ASF infrastructure has blocked a large amount of IPs from GitHub > Actions and as a result, we're seeing random failures in our build. > > This makes github actions worthless for builds since so many of the IPs > have been blocked. Does anyone have an idea how to handle this? Is there an > ASF policy against just using our own server again? We can't really > function as a project if the provided build system times out so much. > > -James >
