I'm trying to install a bunch of C6 involving initially degraded mdadm RAID 1
Anaconda refuses to let me create a RAID 1 array with only one member.
Based on some reading, it seems that I should be able to use kickstart with the PRE scripts to do this. However, after trying for a couple of hours, it doesn't seem that anaconda will allow it, it just boots the created arrays. At best I end up with a list of softraid partitions instead.
Am I trying the impossible or has anybody done this before and can point to a resource how on this can be done?
On Sun, Jun 10, 2012 at 09:34:16PM +0800, Emmanuel Noobadmin wrote:
Based on some reading, it seems that I should be able to use kickstart with the PRE scripts to do this. However, after trying for a couple of hours, it doesn't seem that anaconda will allow it, it just boots the created arrays. At best I end up with a list of softraid partitions instead.
Am I trying the impossible or has anybody done this before and can point to a resource how on this can be done?
Can you do what you want via the GUI installer? Potentially by dropping into a shell to do parts of it (eg build the raid array degreaded, then use the GUI to install onto that). If so, manually do an install and then look at /root/anaconda-ks.cfg to see what file that built out. It might provide the hints you need.
On 6/10/12, Stephen Harris lists@spuddy.org wrote:
Can you do what you want via the GUI installer? Potentially by dropping into a shell to do parts of it (eg build the raid array degreaded, then use the GUI to install onto that). If so, manually do an install and then look at /root/anaconda-ks.cfg to see what file that built out. It might provide the hints you need.
I did do that initially to check it was doable. Unfortunately the anaconda-ks.cfg in that config gave very little clue what was done, there are just 4 relevant lines which look like this #raid / --fstype=ext4 --level=1 --device=md0 raid.253001 raid.253065
Nothing usual, and make sense since I created the md arrays and ananconda simply used them without assembling them itself. The problem is I can't figure how to kickstart to the point where the mds arrays are created successfully.
The command works in shell but somehow in the kickstart process, the created array disappears after ananconda does the "Examining storage device" thing.
On Sun, Jun 10, 2012 at 10:09:28PM +0800, Emmanuel Noobadmin wrote:
#raid / --fstype=ext4 --level=1 --device=md0 raid.253001 raid.253065
Maybe try raid / --fstype=ext4 --level=1 --device=md0 --useexisting Or raid / --fstype=ext4 --device=md0 --useexisting
is I can't figure how to kickstart to the point where the mds arrays are created successfully.
The command works in shell but somehow in the kickstart process, the created array disappears after ananconda does the "Examining storage device" thing.
Why don't you post the necessary fragments of your ks.cfg file and the relevant log messages? At least that way anyone following along won't re-tread your failed paths :-)
On 6/10/12, Stephen Harris lists@spuddy.org wrote:
Maybe try raid / --fstype=ext4 --level=1 --device=md0 --useexisting Or raid / --fstype=ext4 --device=md0 --useexisting Why don't you post the necessary fragments of your ks.cfg file and the relevant log messages? At least that way anyone following along won't re-tread your failed paths :-)
Thanks for the suggestion, I'll try it on a VM and post the cfg file after I'm done fighting the current fire! :)
On 6/10/12, Stephen Harris lists@spuddy.org wrote:
On Sun, Jun 10, 2012 at 10:09:28PM +0800, Emmanuel Noobadmin wrote:
#raid / --fstype=ext4 --level=1 --device=md0 raid.253001 raid.253065
Maybe try raid / --fstype=ext4 --level=1 --device=md0 --useexisting Or raid / --fstype=ext4 --device=md0 --useexisting
is I can't figure how to kickstart to the point where the mds arrays are created successfully.
The command works in shell but somehow in the kickstart process, the created array disappears after ananconda does the "Examining storage device" thing.
Why don't you post the necessary fragments of your ks.cfg file and the relevant log messages? At least that way anyone following along won't re-tread your failed paths :-)
Thanks again for your suggestion, I finally got a chance to try it and wondered why I didn't see the --useexisting option before.
However, now Anaconda complains about partitions required. Fortunately with that additional information, it seems that this is a bug so I don't have to keep banging my head on it.
-
Emmanuel Noobadmin -
Stephen Harris