| Summary: | AB-INT PTEST: tar ptest failure: in 'listing_sparse_files_bigger_than_2^33_B' | ||
|---|---|---|---|
| Product: | [QA/Testing] Package Testing (ptest) | Reporter: | João Marcos Costa <joaomarcos.costa> |
| Component: | ptest | Assignee: | New Comer Bugs <newcomer> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium | CC: | freseenzo, randy.macleod, richard.purdie |
| Version: | unspecified | ||
| Target Milestone: | 6.1 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | AB-INT NEWCOMER | ||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | Don't know | |
|
Description
João Marcos Costa
2026-09-09 13:25:31 UTC
It would be good to check how much free space is left when the tar tests are running. https://autobuilder.yoctoproject.org/valkyrie/#/builders/110/builds/1174/steps/15/logs/stdio qemux86-64-musl-ptest debian13-vk-2 I was able to reproduce this on qemux86-64 with musl. The root cause is
not the host running out of disk: the sparse05 test ("listing sparse
files bigger than 2^33 B") generates a file with 1000 fragments of
10 MiB of real data (~10 GiB), which cannot fit in the 1.4 GB ptest
rootfs. genfile fails with ENOSPC and the test is normally reported as
skipped, but the filesystem stays full for a moment, and when the
autotest bookkeeping (status files, mkdir for test 176) hits that
window the test is reported as FAIL and the next group cannot start.
That timing explains why it looks intermittent.
Patch sent to the list, skipping sparse05 and printing df at the start
of run-ptest as suggested in comment 1:
https://lists.openembedded.org/g/openembedded-core/topic/121269981
With the patch: 200 passed, 46 skipped, 0 failed, no ENOSPC errors.
https://autobuilder.yoctoproject.org/valkyrie/#/builders/73/builds/4569/steps/15/logs/stdio qemux86-64-ptest debian13-vk-1 (Enzo's patch was not included in this one) |