16.01.2017 20:05, Andrei Borzenkov пишет:
16.01.2017 20:01, Andrei Borzenkov пишет:
14.01.2017 21:31, Andrei Borzenkov пишет:
Could you once more try both good and bad cases with udev debugging, but now using
udev.log-priority=debug systemd.log_target=kmsg log_buf_len=16M printk.devkmsg=on
OK, this was more successful. Now we at least see
[ 23.647300] systemd-udevd[623]: IMPORT '/sbin/mdadm --incremental --export /dev/sdc21 --offroot ${DEVLINKS}' /usr/lib/udev/rules.d/64-md-raid-assembly.rules:33 [ 26.224213] systemd-udevd[623]: Process '/sbin/mdadm --incremental --export /dev/sdc21 --offroot ${DEVLINKS}' failed with exit code 1.
for all components.
This proves that startup sequence is correct. What you can try next - insert "-vvv" into mdadm invocation in udev rules; something like
ACTION=="add|change", IMPORT{program}="BINDIR/mdadm -vvv --incremental
s@BINDIR@/sbin@ of course; copied directly from mdadm sources.
--export $devnode --offroot ${DEVLINKS}"
This hopefully will provide some additional output when mdadm fails and this output should be logged by udev. Run bad case with the same kernel parameters. There is no need to run good case as we know where it fails now.
Unfortunately it did not show anything new. mdadm simply silently fails. We can actually see (as debug output in other rules) that device does have correct UUID and MD name as expected by mdadm.conf you shown earlier. So mdadm simply silently fails without any explanation. Sorry, I have to give up. I suggest you open bug report and assign it to mdadm maintainer (nfbrown@suse.com). -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org