| Summary: | AB-INT: core-image-sato.bb:do_testsdk fails on ftpmirror.gnu.org returning 502 Bad Gateway | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [QA/Testing] Functional (self) Testing | Reporter: | Yoann Congal <yoann.congal> | ||||
| Component: | oe-selftest | Assignee: | Yoann Congal <yoann.congal> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | Medium+ | CC: | randy.macleod, tim.orling, yoann.congal | ||||
| Version: | 4.0.32 | ||||||
| Target Milestone: | 4.0.33 | ||||||
| Hardware: | x86 | ||||||
| OS: | Multiple | ||||||
| Whiteboard: | AB-INT | ||||||
| OS type for building Yocto: | --- | Type of Regression: | --- | ||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||
| Attachments: |
|
||||||
|
Description
Yoann Congal
2026-01-15 14:20:17 UTC
This is intermittent: kirkstone qemumips-alt ubuntu2404-vk-3 https://autobuilder.yoctoproject.org/valkyrie/#/builders/13/builds/1417 => failed But, rebuild as kirkstone qemumips-alt alma8-vk-2 https://autobuilder.yoctoproject.org/valkyrie/#/builders/13/builds/1418 => success! That URL is resolved to mirrors, so the actual site is likely random. 502 Bad Gateway is usually something like nginx reverse proxy not being able to load the (often containerized) backend application. Yoann points out this is the mirror URL itself failing before the redirect. Note from bug triage team: * There is code to use DL_DIR instead of downloading it every time on master (and maybe in kirkstone) but I need to check if kirkstone is not missing backports. * Also, master uses downloads.yoctoproject.org instead of upstream servers: oeqa/runtime/buildcpio: Use our own mirror for source ยท openembedded/openembedded-core@038d727 https://github.com/openembedded/openembedded-core/commit/038d7270aef0cccf87d50a117160c58261beb9b9 (RP can help with more context if needed) code is trying to download https://ftpmirror.gnu.org/gnu/cpio/cpio-2.13.tar.gz but it's missing in DL_DIR: ycongal@ubuntu2404-vk-1:~$ ls /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-* -1 /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-2.14.tar.gz /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-2.14.tar.gz.done /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-2.15.tar.gz /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-2.15.tar.gz.done /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-2.8.tar.gz /srv/autobuilder/valkyrie.yocto.io/current_sources/cpio-2.8.tar.gz.done We could either or both: * Maybe switch to cpio-2.14.tar.gz (which is the cpio version of kirkstone, and so, in DL_DIR) to avoid hitting the network * Switch to https://downloads.yoctoproject.org/mirror/sources/cpio-2.13.tar.gz to avoid the unreliable ftpmirror.gnu.org server Note: master does not use downloads.yoctoproject.org everywhere: meta/lib/oeqa/sdk/cases/autotools.py: tarball = self.fetch(testdir, self.td["DL_DIR"], "https://ftpmirror.gnu.org/gnu/cpio/cpio-2.15.tar.gz") meta/lib/oeqa/selftest/cases/meta_ide.py: "https://ftpmirror.gnu.org/gnu/cpio/cpio-2.15.tar.gz", I conclude that it does not hit this bug because cpio-2.15.tar.gz is always in DL_DIR. So, I'll send a patch for kirkstone to align the test archive with the cpio recipe to ensure it is in DL_DIR. Backported fix merged on kirkstone: oeqa: Use 2.14 release of cpio instead of 2.13 https://git.openembedded.org/openembedded-core/commit/?h=kirkstone&id=64d56cf416b31ae92438deefe4028402120ed998 => Closing Note: If that ever happen again, we can still switch to yp.org mirrors. |