Summary: | sysutils/e2fsprogs: build failure due to failing tests | ||
---|---|---|---|
Product: | Ports & Packages | Reporter: | O. Hartmann <ohartmann> |
Component: | Individual Port(s) | Assignee: | Matthias Andree <mandree> |
Status: | Closed Feedback Timeout | ||
Severity: | Affects Many People | Flags: | mandree:
maintainer-feedback+
|
Priority: | --- | ||
Version: | Latest | ||
Hardware: | Any | ||
OS: | Any |
Description
O. Hartmann
2019-10-21 08:29:04 UTC
Oliver, that's indeed mysterious and I wonder what's common to all of the systems you oversee - is there a certain configuration that you add? src.conf, rc.conf, make.conf? What are your file system types, what is timestamp granularity, how do you synchronize time between file server and poudriere host? Is there anything that makes time jumpy or otherwise non-continuous? Networked file systems that cache time attributes that make the system time-travel somehow? Please upload or send me: 1. logs where all file systems that you use have high-resolution timestamps and where ntp doesn't make time jump. -STABLE or -RELEASE strongly preferred. 2. a xz'd or zstd'd log of a build where the host is a supported FreeBSD RELEASE, not a moving target such as -STABLE or -CURRENT, and where you have a fully-updated jail that is also a RELEASE version that is not newer than the host so we can start isolating the issue with fewer variables. 21 failures pretty much looks either like time stamping issues (where the file server has a different time than the host that mounts exported file systems) or else like a toolchain or world issue rather than a port issue. I thought we'd shaken out the last reproducible non-tier-1 bug in self-tests with 1.45.3_2 on 2019-07-30, by importing https://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git/commit/?id=a368e0cbfb33d3050dcccccf0bf5a5539d3dac39 into r507631. Which compiler are you using? |