[Bug 1161331] New: [y2-storage-ng] Storage proposal: infinite loop when installing on top of a software RAID
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331 Bug ID: 1161331 Summary: [y2-storage-ng] Storage proposal: infinite loop when installing on top of a software RAID Classification: openSUSE Product: openSUSE Tumbleweed Version: Current Hardware: Other OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: YaST2 Assignee: yast2-maintainers@suse.de Reporter: ancor@suse.com QA Contact: jsrain@suse.com Found By: --- Blocker: --- When a software RAID is completely empty (not formatted and not partitioned), it will be considered as a candidate device for the storage proposal. But currently that can lead to an infinite loop in such situations. I will know link to a testcase that proves that situation exists. -- You are receiving this mail because: You are on the CC list for the bug.
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c1
--- Comment #1 from Ancor Gonzalez Sosa
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
Ancor Gonzalez Sosa
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c2
Ancor Gonzalez Sosa
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c3
Ancor Gonzalez Sosa
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c4
--- Comment #4 from Ancor Gonzalez Sosa
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c5
Shivani Changela
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c6
Ancor Gonzalez Sosa
So my question is: do my explanations fit your expectations about how the proposal should work on top of "virtual disks"?
Setting the needinfo on Shivani Changela and Nanda Kishore again, for them to reply to my questions about expectations and booting from software RAIDs. Talking about expectations, let me clarify two points. First, in the logs attached by Shivani Changela I can see these systems use EFI to boot. So everytime I say "partition of type bios_boot" in comment#3 I should had actually said "an ESP (EFI) partition". All other considerations stay exactly as they are. That being said, in the mentioned logs uploaded by Shivani Changela I can see this: A disk "/dev/sda" with a total size of 220 GiB and GPT partition table It contains a single "/dev/sda1" partition of 100 GiB A disk "/dev/sdb" with a total size of 910 GiB and GPT partition table It contains no partitions A disk "/dev/sdc" with a total size of 440 GiB and GPT partition table It contains a single "/dev/sdc1" partition of 100 GiB A disk "/dev/sdd" fully empty (not even partition table) A software RAID1 called "/dev/md/VirtualDisk01" formed by sda1 and sdc1 It contains a GPT partition table with: - An ESP (EFI) partition - Two Linux partitions formatted with XFS - A swap partition According to what I explained in comment#3, we (the YaST Team) wouldn't expect such system to be able to boot, since there is no ESP partition allocated in any real disk. The only ESP partition is inside a software RAID that is backed by partitions. So my question here is: does such system boot? So far, we had no evidence of S130/S140 controllers being able to boot from a software RAID if the needed partitions were not present in real physical disks. If you can prove that works, we would need to reconsider how the proposal works in those cases. -- You are receiving this mail because: You are on the CC list for the bug.
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
Steffen Winterfeldt
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c7
NANDA KISHORE
(In reply to Ancor Gonzalez Sosa from comment #3)
So my question is: do my explanations fit your expectations about how the proposal should work on top of "virtual disks"?
Setting the needinfo on Shivani Changela and Nanda Kishore again, for them to reply to my questions about expectations and booting from software RAIDs.
Talking about expectations, let me clarify two points.
First, in the logs attached by Shivani Changela I can see these systems use EFI to boot. So everytime I say "partition of type bios_boot" in comment#3 I should had actually said "an ESP (EFI) partition". All other considerations stay exactly as they are.
That being said, in the mentioned logs uploaded by Shivani Changela I can see this:
A disk "/dev/sda" with a total size of 220 GiB and GPT partition table It contains a single "/dev/sda1" partition of 100 GiB
A disk "/dev/sdb" with a total size of 910 GiB and GPT partition table It contains no partitions
A disk "/dev/sdc" with a total size of 440 GiB and GPT partition table It contains a single "/dev/sdc1" partition of 100 GiB
A disk "/dev/sdd" fully empty (not even partition table)
A software RAID1 called "/dev/md/VirtualDisk01" formed by sda1 and sdc1 It contains a GPT partition table with: - An ESP (EFI) partition - Two Linux partitions formatted with XFS - A swap partition
According to what I explained in comment#3, we (the YaST Team) wouldn't expect such system to be able to boot, since there is no ESP partition allocated in any real disk. The only ESP partition is inside a software RAID that is backed by partitions.
So my question here is: does such system boot? So far, we had no evidence of S130/S140 controllers being able to boot from a software RAID if the needed partitions were not present in real physical disks. If you can prove that works, we would need to reconsider how the proposal works in those cases.
Ancor, till date, S130/S140/S150 was always working by booting from RAID1 ESP. On SLES15-SP1 is automatically creating ESP on the MDRAID Partition unlike the way you have mentioned in Comment #3 i.e. ESP should be on a physical drive/partition. Attaching the supporting logs from SLES15-SP1 installation(since SP2 we have infinite loop bug):y2logs, suggested paritioning screenshot, post os boot logs. -- You are receiving this mail because: You are on the CC list for the bug.
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c8
--- Comment #8 from NANDA KISHORE
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c9
--- Comment #9 from NANDA KISHORE
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c10
--- Comment #10 from NANDA KISHORE
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c11
--- Comment #11 from Ancor Gonzalez Sosa
This is the unit test proving the bug exists https://github.com/yast/yast-storage-ng/pull/1027
I closed that in favor of https://github.com/yast/yast-storage-ng/pull/1043 The new pull request is based on the SLE-15-SP1 branch and proves the behavior of that release is exactly the one described by Nanda Kishore. So the goal here is not only to avoid the infinite loop but to make sure the same test run with the same result in the SLE-15-SP2 codebase. -- You are receiving this mail because: You are on the CC list for the bug.
![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
http://bugzilla.suse.com/show_bug.cgi?id=1161331
http://bugzilla.suse.com/show_bug.cgi?id=1161331#c12
Ancor Gonzalez Sosa
participants (1)
-
bugzilla_noreply@novell.com