Bug 761278
| Summary: | Fedora16 installer fails when drive is already partitioned - system with multipath | ||
|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | IBM Bug Proxy <bugproxy> |
| Component: | anaconda | Assignee: | Anaconda Maintenance Team <anaconda-maint-list> |
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
| Severity: | urgent | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 16 | CC: | anaconda-maint-list, dcantrell, jkachuck, jonathan, pknirsch, vanmeeuwen+fedora, wgomerin |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | ppc64 | ||
| OS: | All | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2012-07-19 21:20:54 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 803331 | ||
| Attachments: | |||
|
Description
IBM Bug Proxy
2011-12-07 23:01:13 UTC
Created attachment 542242 [details]
anaconda.log
Created attachment 542243 [details]
program.log
Created attachment 542244 [details]
storage.log
Created attachment 542245 [details]
attached picture
The basic problem is that the code in the devicetree to handle scanning and activating multipath devices does not work if the mpaths are already active when the devicetree scans the system. Since anaconda scans for devices at least twice during install, only the first scan is successful as far as mpaths go. One possible solution that occurs to me is to ignore dm devices when iterating self.topology.devices_iter in DeviceTree._populate. They'll get scanned in the subsequent loops after the mpath-specific set has been done. Created attachment 544725 [details] Test fix for bug 77158 ------- Comment (attachment only) From dlshaw.com 2011-12-09 16:52 EDT------- ------- Comment From dlshaw.com 2011-12-09 17:01 EDT------- Mark Hamzy provided me with a fix to test on my Juno IOC-L hardware that has a SAS controller with multipath (mpath). I've attached the 77158-1.img file to this bug as well. This patch fixed the problem where we couldn't install Fedora16 on a system with SAS and mpath. I also regression tested this on a system with SAS and no mpath successfully. However, there is one problem left to resolve on a SAS with mpath. Once Fedora starts the VNC installation the user will choose the following options: Language Keyboard Layout Storage Device After selecting Basic storage device the installer will examine the system's storage device(s) and should detect any existing Fedora installations. If it detects an existing installation the installer should display a menu that notifies the user that it has detected an existing installation and asks the user what they wish to do. Fresh Install or Upgrade existing installation. This behaves correctly on SAS without mpath systems but for systems that have SAS and mpath the installer fails to detect any existing installations and doesn't display the menu letting the user know there were existing installation(s) detected. Instead it goes staight to the next menu which I think is the hostname of the system. I'm going to attach a screen shot of the installation menu screen that should show up when an existing installation has been detected. ------- Comment From dlshaw.com 2011-12-09 17:04 EDT------- I captured fdisk -l output from a Juno IOC-L that has SAS without mpath in case its helpful to figure this out. sh-4.2# fdisk -l Disk identifier: 0x64cd0800 Disk identifier: 0xff920500 Disk /dev/sda: 73.4 GB, 73407488000 bytes 128 heads, 32 sectors/track, 35003 cylinders, total 143374000 sectors Disk identifier: 0x82bc0000 /dev/sda3 1034240 143372287 71169024 8e Linux LVM Disk /dev/sdd: 73.4 GB, 73407488000 bytes 128 heads, 32 sectors/track, 35003 cylinders, total 143374000 sectors Disk identifier: 0x5eee0100 ------- Comment From hamzy.com 2011-12-09 17:09 EDT------- dlehman, the fix still does not detect an existing system. Do you know what might be wrong there? Why does it detect the disks a number of times? Is there a way to deactivate the multipath devices before the first scan occurs? Created attachment 544746 [details]
Existing installation detected menu
------- Comment (attachment only) From dlshaw.com 2011-12-09 17:02 EDT-------
Created attachment 544747 [details]
Patch as suggested by dlehman
------- Comment (attachment only) From hamzy.com 2011-12-09 17:06 EDT-------
I believe you must choose advanced storage to use mpath. I would have to see the storage.log to see what specifically is not happening. Created attachment 546406 [details]
fc16 beta rc 6 - doesn't detect existing OS with mpath device
------- Comment on attachment From dlshaw.com 2011-12-13 16:23 EDT-------
Installer doesn't detect existing OS. I picked "Specialized Storage Devices" and selected the two mpath devices listed. This is with the default Fedora16 Beta RC6 installer without updates.
Created attachment 546407 [details]
fc16 beta rc 6 - doesn't detect existing OS with mpath device w/ Mark's 77158-1.img update file
------- Comment on attachment From dlshaw.com 2011-12-13 16:24 EDT-------
Installer doesn't detect existing OS. I picked "Specialized Storage Devices" and selected the two mpath devices listed. This is with the default Fedora16 Beta RC6 installer with Mark's 77158-1.img update file.
------- Comment From hamzy.com 2011-12-13 16:34 EDT------- Any ideas, dlehman, what else might be wrong? ------- Comment From hamzy.com 2011-12-20 17:02 EDT-------
If I look at the machine when it is running the installed Fedora 16 partition, the multipath code seems to be able to parse the output correctly.
[root@pfdioc03b ~]# yum install -y anaconda device-mapper-multipath
[root@pfdioc03b ~]# multipath -ll
Dec 20 15:18:57 | DM multipath kernel driver not loaded
[root@pfdioc03b ~]# modprobe dm-multipath
[ 874.000422] device-mapper: multipath: version 1.3.0 loaded
[root@pfdioc03b ~]# multipath -ll
[root@pfdioc03b ~]# multipath -d
create: mpatha (35000c50017b69adf) undef IBM,ST9146852SS
size=137G features='0' hwhandler='0' wp=undef
|-+- policy='round-robin 0' prio=1 status=undef
| `- 0:0:4:0 sda 8:0 undef ready running
`-+- policy='round-robin 0' prio=1 status=undef
`- 1:0:4:0 sdc 8:32 undef ready running
create: mpathb (35000c5002328f6a7) undef IBM,ST9146852SS
size=137G features='0' hwhandler='0' wp=undef
|-+- policy='round-robin 0' prio=1 status=undef
| `- 0:0:5:0 sdb 8:16 undef ready running
`-+- policy='round-robin 0' prio=1 status=undef
`- 1:0:5:0 sdd 8:48 undef ready running
[root@pfdioc03b ~]# python
>>> from pyanaconda.storage.devicelibs import mpath
>>> from pyanaconda import iutil
>>> output = iutil.execWithCapture("multipath", ["-d",])
>>> mpaths = mpath.parseMultipathOutput(output)
>>> mpaths
{'mpathb': ['sdb', 'sdd'], 'mpatha': ['sda', 'sdc']}
------- Comment From hamzy.com 2011-12-20 18:46 EDT------- Submitted the patch to the anaconda development mailing list. https://www.redhat.com/archives/anaconda-devel-list/2011-December/msg00108.html dlehman recommends that we close this defect and open a new one for the "not detecting an existing installation" problem. ------- Comment From clnperez.com 2012-02-27 12:17 EDT------- (In reply to comment #22) > Submitted the patch to the anaconda development mailing list. > > https://www.redhat.com/archives/anaconda-devel-list/2011-December/msg00108.html > > dlehman recommends that we close this defect and open a new one for the "not > detecting an existing installation" problem. Mark, It looks like that's been done (Bug 77584), so I'm going to close this one. Per IBM this bz should have been closed (Comment #17) ------- Comment From clnperez.com 2012-06-11 18:56 EDT------- Yes, it has been closed on our side. Thanks. Closing due to comment #19 |