ci: intermittent failure in Windows CI

fanquake commented at 12:26 pm on April 3, 2026: member

https://github.com/bitcoin/bitcoin/actions/runs/23945053179/job/69839333752?pr=34978#step:8:587:

 0Downloading https://github.com/boostorg/mpl/archive/boost-1.88.0.tar.gz -> boostorg-mpl-boost-1.88.0.tar.gz
 1error: https://github.com/boostorg/mpl/archive/boost-1.88.0.tar.gz: failed: status code 502
 2note: If you are using a proxy, please ensure your proxy settings are correct.
 3Possible causes are:
 41. You are actually using an HTTP proxy, but setting HTTPS_PROXY variable to `https://address:port`.
 5This is not correct, because `https://` prefix claims the proxy is an HTTPS proxy, while your proxy (v2ray, shadowsocksr, etc...) is an HTTP proxy.
 6Try setting `http://address:port` to both HTTP_PROXY and HTTPS_PROXY instead.
 72. If you are using Windows, vcpkg will automatically use your Windows IE Proxy Settings set by your proxy software. See: https://github.com/microsoft/vcpkg-tool/pull/77
 8The value set by your proxy might be wrong, or have same `https://` prefix issue.
 93. Your proxy's remote server is out of service.
10If you believe this is not a temporary download server failure and vcpkg needs to be changed to download this file from a different location, please submit an issue to https://github.com/Microsoft/vcpkg/issues
11CMake Error at scripts/cmake/vcpkg_download_distfile.cmake:136 (message):
12  Download failed, halting portfile.
13Call Stack (most recent call first):
14  scripts/cmake/vcpkg_from_github.cmake:120 (vcpkg_download_distfile)
15  C:/Users/runneradmin/AppData/Local/vcpkg/registries/git-trees/6cdd95078ed4dc9bd700884cfc132dca24a99013/portfile.cmake:3 (vcpkg_from_github)
16  scripts/ports.cmake:206 (include)

We have recently added logic, to try and prevent these kinds of intermittent failures, by retrying. Seen here: https://github.com/bitcoin/bitcoin/actions/runs/23945053179/job/69839333752?pr=34978#step:8:738)

Maybe we should retry more aggressively?

fanquake added the label Windows on Apr 3, 2026

fanquake added the label CI failed on Apr 3, 2026

JeremyRubin commented at 8:47 pm on April 3, 2026: contributor

maybe they should just get downloaded once and put multiple somewheres under bitcoin control that is easy to retrieve, rather than through a github based archive?

Rather than retrying against the same endpoints…

fanquake commented at 1:19 am on April 4, 2026: member

Sure. One issue is that I don’t think we know what files are needed, it’s whatever vcpkg decides to download. cc @hodlinator @hebasto Any thoughts?

maflcko commented at 8:47 am on April 7, 2026: member

There could be a second re-try after 50 seconds or so, but with a 502 error, it is unclear how long it will take to come back.

I think the GHA VMs are also randomly disappearing during a run, leading to a 2 hour timeout, which happens more often than this issue, so it may also be fine to accept this issue as-is and deal with it as part of dealing with the other GHA VM issues/timeouts. That is, by manually re-running it on network errors or other timeouts.

hodlinator commented at 9:03 am on April 7, 2026: contributor

(I’m not familiar with vcpkg proxying/caching). Sad that GitHub itself cannot be relied upon. A week or so ago I had to change my NixOS setup to use a GitHub authentication token to avoid “Resource temporarily unavailable” errors. But that’s different from 502 Bad Gateway.

ci: intermittent failure in Windows CI #34996