| Summary: | Dracut does not add mdadm to initramfs when needed | ||
|---|---|---|---|
| Product: | Mageia | Reporter: | Anssi Hannula <anssi.hannula> |
| Component: | RPM Packages | Assignee: | Colin Guthrie <mageia> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | critical | ||
| Priority: | release_blocker | ||
| Version: | 3 | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Source RPM: | dracut | CVE: | |
| Status comment: | |||
Fixed by reverting a bad patch I added :( http://svnweb.mageia.org/packages?view=revision&revision=412382 Already pushed and available. Status:
NEW =>
RESOLVED |
mdadm is missing from initramfs since mga2->mga3 upgrade, reproducable after upgrade as well so not a transient issue. This causes a failure to boot on an mdadm RAID system. 18:16:22 < Anssi> dracut seems to think md is not needed... but why, that is a bit more difficult to figure out... 18:18:11 < Anssi> and the code seems like it can't possibly work... 90mdraid/module-setup.sh::check() calls for_each_host_dev_and_slaves_all() giving check_mdraid() as a callback function... but the callback expects 2 parameter while it actually gets called with 1 18:36:07 < Anssi> coling: applied this http://git.kernel.org/cgit/boot/dracut/dracut.git/commit/?id=d351541ee625f8a1b42446448214f200f19c2631 which in modules.d/90mdraid/module-setup.sh replaces the call to for_each_host_dev_and_slaves_all with for_each_host_dev_fs among other things, and mdadm gets now added correctly If it turns out that this does not affect all root-on-mdRAID systems, priority can be lowered from release-blocker.