12.01.2017 00:35, Istvan Gabor пишет:
On Wed, 11 Jan 2017 06:58:21 +0300, Andrei Borzenkov wrote:
10.01.2017 22:32, Istvan Gabor пишет: ...
Upload output of "journalctl -b" somewhere, e.g. http://susepaste.org/
I tried to upload the output from journalctl -b to susepaste.org but the site haven't accepted it. I guess it's too much to paste in. The whole log file is ~1 MB and it contains 11678 lines. I have sent the log to your email too, I don't know if you've received it.
I did; as I answered in another mail, udev debug output was not that useful and there is nothing standing out between good and bad cases. Also some logs are definitely lost.
Hi Andrei,
Was the new log file I sent this morning useful?
Not really. It does show that udevadm trigger is run second time after switch to real root, so this can be ruled out.
I have tried openSUSE 13.2 and it assembles the arrays correctly at boot.
Well, it used completely different approach, so it not comparable. As I mentioned, one problem is that logs are lost (apparently, journal cannot cope with this amount or it is started too late). Log for working case does show traces of udev event(s) that trigger raid assembly - but only second part of them. Unfortunately kernel appears to ratelimit messages as well; fortunately Leap kernel appears to include recent patch to control it on kernel command line. 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 The later parameter disables ratelimiting in kernel. We should now hopefully get full log and it is enough to just send output of "dmesg" (as everything must be now in kernel buffer). Please compress files, thet are expected to be rather large. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org