https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c Xiyuan Liu changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|bnc-team-screening@forge.pr |werner@suse.com |ovo.novell.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c Dr. Werner Fink changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |cus@fazekas.hu -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c4 Marton Balint changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|cus@fazekas.hu | --- Comment #4 from Marton Balint 2013-11-16 03:16:45 UTC --- Created an attachment (id=567743) --> (http://bugzilla.novell.com/attachment.cgi?id=567743) Output of disk.sh script My /etc/init.d/halt.local is empty. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c6 Marton Balint changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #567743|0 |1 is obsolete| | --- Comment #6 from Marton Balint 2013-11-16 03:24:52 UTC --- Created an attachment (id=567745) --> (http://bugzilla.novell.com/attachment.cgi?id=567745) Output of fixed disk.sh script -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c7 Dr. Werner Fink changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |werner@suse.com AssignedTo|werner@suse.com |systemd-maintainers@suse.de --- Comment #7 from Dr. Werner Fink 2013-11-18 15:12:50 UTC --- Hmmm ... on the first sight I'd like to guess that all four disks as handled by the kernel, that fushing the disk cache as well as the meta data of the file systems. No need to add my old code from hddown.c from old sysvinit. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c9 --- Comment #9 from Marton Balint 2013-11-18 19:21:36 UTC --- Created an attachment (id=567923) --> (http://bugzilla.novell.com/attachment.cgi?id=567923) /etc/fstab -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c11 Marton Balint changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|cus@fazekas.hu | --- Comment #11 from Marton Balint 2013-11-18 19:28:36 UTC --- Created an attachment (id=567925) --> (http://bugzilla.novell.com/attachment.cgi?id=567925) Shutdown image After I see the storage is finalized message, the disks start to spin down, then at least one starts to spin up. Then I see the very last "powering off" message and the system powers off. It seems my system is not logging to /var/log/boot.log. Do I have to enable it somewhere? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c13 Dr. Werner Fink changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |cus@fazekas.hu --- Comment #13 from Dr. Werner Fink 2013-11-19 08:21:58 UTC ---

From source code of systemd I do *not* see how systemd can influence the kernels behaviour on spin up/down a disk at poweroff:

if (retries >= FINALIZE_ATTEMPTS) log_error("Too many iterations, giving up."); else log_info("Storage is finalized."); arguments[0] = NULL; arguments[1] = argv[1]; arguments[2] = NULL; execute_directory(SYSTEM_SHUTDOWN_PATH, NULL, arguments); [... initrd part skipped ...] /* The kernel will automaticall flush ATA disks and suchlike * on reboot(), but the file systems need to be synce'd * explicitly in advance. So let's do this here, but not * needlessly slow down containers. */ if (!in_container) sync(); [ ... kexec part skipped ..] reboot(cmd); only the line with YSTEM_SHUTDOWN_PATH lead .. if expanded ... to the directory /usr/lib/systemd/system-shutdown/ please list the content of this directory ... I guess we will find the script mdadm.shutdown but maybe there is more than this one? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c15 Marton Balint changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|cus@fazekas.hu | --- Comment #15 from Marton Balint 2013-11-19 22:09:35 UTC --- /usr/lib/systemd/system-shutdown/ only contains the mdadm.shutdown script. I've made some further tests, and it seems my original report was not accurate, the issue is not always reproducible, and it can also happen during a halt (not only during poweroff). I have not been able to reproduce the issue when the /usr/lib/systemd/system-shutdown/ directory was empty, but maybe it's just bad luck again. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c17 --- Comment #17 from Marton Balint 2013-11-20 20:42:03 UTC --- Maybe I misread the man page, but it says that this only matters for arrays with external metadata, but all my arrays are using internal metadata. I tried commenting out the mdadm command, and the issue still happens with an empty shell script, so I don't think the mdadm command itself is the cause. On the other hand, I still have not been able to reproduce the issue with an empty systemd-shutdown directory, I will keep on trying... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c19 Dr. Werner Fink changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO CC| |stefan.fent@suse.com InfoProvider| |nfbrown@suse.com --- Comment #19 from Dr. Werner Fink 2013-11-21 09:56:05 UTC --- (In reply to comment #17) AFAICS from your /etc/fstab and /proc/mdstat the directory /usr/lib/systemd/system-shutdown is located on md0 aka root file sysem / ... what happens after the storage is finalized? Does the kernel spin down the disks afterwards but before the shell scripts below /usr/lib/systemd/system-shutdown will be read for execution? Please show the output of find /sys/block -name '[sh]d[a-z]' | while read disk ; do disk=${disk##*/} hdparm -C -B /dev/${disk} done currently I'm only aware that Seagate disks may have a auto powersaving mode which could be disabled by find /sys/block -name '[sh]d[a-z]' | while read disk ; do disk=${disk##*/} hdparm -Z /dev/${disk} done but AFAICS you're using Samsung disks -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c21 systemd maintainers changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |cus@fazekas.hu --- Comment #21 from systemd maintainers 2013-11-22 09:07:07 UTC --- (In reply to comment #20)

I beleive the kernel is powering off the disks when systemd calls reboot(cmd) and not sooner.

Hmmm ... this is not what you've said in comment #17 <cite> I tried commenting out the mdadm command, and the issue still happens with an empty shell script, so I don't think the mdadm command itself is the cause. On the other hand, I still have not been able to reproduce the issue with an empty systemd-shutdown directory, I will keep on trying... </cite> .. now what is the correct behaviour? If the spin down and spin up really happens even with empty systemd-shutdown directory, then this is a kernel related problem. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c23 Dr. Werner Fink changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |cus@fazekas.hu --- Comment #23 from Dr. Werner Fink 2013-11-22 10:30:51 UTC --- Try add a script with three lines #!/bin/bash sleep 5 echo -en '\a' and make it executable. Now what does happen? Beside this can you identify *which* disk does spin up? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c25 Marton Balint changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|cus@fazekas.hu | --- Comment #25 from Marton Balint 2013-12-01 16:19:18 UTC --- With the script, after the storage is finalized line is printed and before the disks start to spin down, there is a 5 second delay. There is no beep, I think systemd swallows the output of the scripts. I could not identify the disk so far. The order of the disk spindown is sdd, sdc, sdb, sda. Since I hear the spin up sound before the last disk starts spinning down, it is safe to say it is NOT sda. Most probably it is either sdd or sdc. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c27 Takashi Iwai changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO CC| |tiwai@suse.com InfoProvider| |cus@fazekas.hu --- Comment #27 from Takashi Iwai 2014-03-19 17:24:35 UTC --- it's difficult to know what's happening there without outputs and logs (and it's difficult to get such in this case). One thing to be worth would be to try the newer upstream kernels. Does this problem remain with the latest 3.13.x kernel and 3.14-rc kernel? They can be found in OBS Kernel:stable and Kernel:HEAD repos. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=849692 https://bugzilla.novell.com/show_bug.cgi?id=849692#c29 Takashi Iwai changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |cus@fazekas.hu --- Comment #29 from Takashi Iwai 2014-03-23 10:10:28 UTC --- OK, thanks. If it's reproduced with the upstream kernel, it's better to be reported to upstream, too. Care to report it on bugzilla.kernel.org? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

http://bugzilla.novell.com/show_bug.cgi?id=849692 --- Comment #31 from Marton Balint --- Feel free to close this, I can no longer reproduce it. Unfortunately a lot of has changed since I my original report, (new motherboard, new opensuse, removal of Samsung SP2504C from the machine), so I am not sure which one fixed it, but probably it was the Samsung SP2504C which behaved strangely in the first place. -- You are receiving this mail because: You are on the CC list for the bug.