2016-12-11 13:59 UTC

View Issue Details Jump to Notes ]
IDProjectCategoryView StatusLast Update
0002940CentOS-5anacondapublic2008-07-02 14:41
Reporterelias1884 
PrioritynormalSeveritycrashReproducibilityalways
StatusnewResolutionopen 
Product Version5.2 
Target VersionFixed in Version 
Summary0002940: Anaconda aborts saying it can not write to a RAID1 member partition, which has not been altered nor used for the installation
DescriptionI am trying to install Centos5.2 on a System which already has a Debian Etch installed.

Partition / Raid Layout
md0 sda/b1 Etch /boot
md1 sda/b2 /boot for my new CentOS 5.2
md2 sda/b3 Other CentOS 5.2 installation's /boot
md3 sda/b5 Etch /
md4 sda/b6 / form my new CentOS 5.2
md5 sda/b7 Other CentOS 5.2 installation's /
md6 sda/b8 LVM2 (only used for Xen guests)

The other CentOS 5.2 system has not been installed with Anaconda but is a copy of a Xen guest, in case you are wondering.

Now, when I only alter md2 and md4 in anaconda to format as ext2 and ext3 as /boot and /, I get an error message complaining that changes to /dev/sda5 could not be read. WTF is it trying to do with sda5 you are asking yourself? So do I.

I have originally partitioned the drive with Debian Etch installer.

I get this every time I try to install!

I have attached anaconda's debug log.
TagsNo tags attached.
Attached Files

-Relationships
+Relationships

-Notes

~0007539

elias1884 (reporter)

Disk /dev/sda: 400.0 GB, 400088457216 bytes
255 heads, 63 sectors/track, 48641 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

   Device Boot Start End Blocks Id System
/dev/sda1 * 1 26 208844+ fd Linux raid autodetect
/dev/sda2 * 27 52 208845 83 Linux
/dev/sda3 * 53 78 208845 fd Linux raid autodetect
/dev/sda4 79 48641 390082297+ 85 Linux extended
/dev/sda5 79 1384 10490444+ fd Linux raid autodetect
/dev/sda6 1385 2690 10490444+ fd Linux raid autodetect
/dev/sda7 2691 3996 10490444+ fd Linux raid autodetect
/dev/sda8 3997 48641 358610931 fd Linux raid autodetect

Disk /dev/sdb: 400.0 GB, 400088457216 bytes
255 heads, 63 sectors/track, 48641 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

   Device Boot Start End Blocks Id System
/dev/sdb1 * 1 26 208844+ fd Linux raid autodetect
/dev/sdb2 * 27 52 208845 83 Linux
/dev/sdb3 * 53 78 208845 fd Linux raid autodetect
/dev/sdb4 79 48641 390082297+ 85 Linux extended
/dev/sdb5 79 1384 10490444+ fd Linux raid autodetect
/dev/sdb6 1385 2690 10490444+ fd Linux raid autodetect
/dev/sdb7 2691 3996 10490444+ fd Linux raid autodetect
/dev/sdb8 3997 48641 358610931 fd Linux raid autodetect

~0007542

elias1884 (reporter)

That is the error message anaconda gave me:
Error informing the kernel about modifications to partition /dev/sda5 -- Device or resource busy. This means Linux won't know about any changes you made to /dev/sda5 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.

Clicking OK did exit anaconda

~0007543

elias1884 (reporter)

Could be my anaconda problem is coming from here: http://fedoraproject.org/wiki/Bugs/F7Common#Partition_table_claimed_to_be_invalid_on_previously_working_drive

~0007544

elias1884 (reporter)

Fedora 9 does show the same error, but hitting "ignore" lets me finish the installation as opposed to Centos 5.2, which does quit the installation process even when hitting ignore!

~0007545

elias1884 (reporter)

Centos throws an exception when hitting ignore

~0007547

timverhoeven (developer)

Could you try the install with the option given in that Fedora wiki article you linked to ? It might fix it.

~0007548

elias1884 (reporter)

Unfortunately I can't do that. I have a server at Hetzner.de and am using their VNC installation routine, which does not allow me to pass any kernel params.

~0007550

elias1884 (reporter)

I very much doubt that this has anything to do with HPA, as HPA generally concerns the end of a disk. /dev/sda5 however is in the middle of the drive (first extended partition) therefore even if there was a HPA, it should not concern /dev/sda5.

Also my Server came with blank disks and as this is no-name hardware there is not DELL or IBM stuff in HPA.

What I found out however is, that it seems anaconda treats /dev/sda5 as a disk instead of as a partition. It tries to reload the changes made to its partition table as it seems.

Also, what appeared to be very interesting was the fact that on the console I tried to check the raid superblocks when the error message popped up (before clicking ignore). It appeared as non of the extended partitions on sda was known by the system = dev or file not found. While I could access everything on sdb.

Debian Etch however can boot and assemble all Raids without any problems. I really have no idea what is going on, but it does not seem to be HPA. Unfortunately Etch's dmesg does not show any HPA stuff either, so I can not confirm with 100% accuracy that there is not HPA. But I am 99,99% sure there is none.

Why should a blank disk, bought by bulk have a HPA set?
+Notes

-Issue History
Date Modified Username Field Change
2008-07-01 14:27 elias1884 New Issue
2008-07-01 14:27 elias1884 File Added: centos5.error
2008-07-01 14:49 elias1884 Note Added: 0007539
2008-07-01 21:25 elias1884 Note Added: 0007542
2008-07-01 21:33 elias1884 Note Added: 0007543
2008-07-01 23:23 elias1884 Note Added: 0007544
2008-07-01 23:53 elias1884 Note Added: 0007545
2008-07-02 08:03 timverhoeven Note Added: 0007547
2008-07-02 12:35 elias1884 Note Added: 0007548
2008-07-02 14:41 elias1884 Note Added: 0007550
+Issue History