issue with setup-storage, squeeze, and SW Raid1
dave at socialstrata.com
Wed Dec 22 03:50:11 CET 2010
I've been tasked with updating our Lenny FAI server to run on and install Squeeze. The clients that we'll be installing to are a mix of servers, some of which are IDE, 2 drives, to be running Raid1.
Thanks to the tireless efforts of Michael Tautschnig we've had this configuration running for a time on Lenny with no issues.
I reinstalled the fai server on Squeeze, installed 4.0~beta2+experimental48 and copied over all of our working configs from the Lenny 4.0~beta2+experimental45 machine.
Upon trying to install we found several problems.
Our /dev/hdx drives were being identified as /dev/sd(x), confusing both us and setup storage. I made edits to my disk_config file to work around this and restarted, the process went all the way through, with some fairly ugly entries in the error log about disk performance, which was to be expected.
What we did not expect was that upon rebooting the server could no longer access our hard drives! The bios would show the two drives, begin a S.M.A.R.T. check then freeze, totally. I thought it odd that two drives would go bad at the same time and took two new IDE drives out of new packaging and dried it again. Same thing!
I understand that the afore description is probably pretty terrible, so here is a screen capture of the console. There is not much to see, hence the description not being all that good. http://www.web-performance.net/IMG_20101217_134528.jpg What should happen is that the computer should proceed from here into PXE and then start the tftp process, however it will not pass the point seen in the image. Obviously there are no logs available at this point in the process.
Now, as I refused to believe that 4 drives, 2 of which were brand new, all died at once I took them over to an old Winxp PC we have laying around and plugged one of these 4 drives into it. The machine booted fine. I the Windows Disk management tool I was able to see and remove the partitions that sertup-storage created on this drive. I repeated that process for all 4 drives. I then carried the drives back over to the client server that I was trying to use, and put 2 of these discs back in. The machine detected them, had no issues with them, went past the point where they had been locking the server, and into PXE, the tftp, and then FAI as they should.
I repeated this with two different client server chassis (same model) and 2 different Winxp PC computers. Absolutely repeatable each time.
Finally, I downloaded the fai-client_3.3.5_all.deb, extracted the setup-storage script and the perl modules, put them into my nfsroot and started one of the clients up. Perfect installation. No problem. Client server reboots and has no disc or partition issues at all.
Squeeze is still identifying my IDE drives as /dev/sd(x) for instance in fdisk -l, but hey, it works...... ;)
Whatever the issue is, I think it is has to be in the experimental 48 build of setup-storage. I'm just not savvy enough to figure out what the problem is as there are no errors logged that I can see. I have a tarball of the entire log directory from the last installation that couldn't boot, on the 48 build. Perhaps there is something in there to look at? If someone wants it I can send it to him as I don't think I can (or should) attach it in an email to the group here.
Customer Advocate, Social Strata, Inc.
Online Community HQ Since 1996
Hoop.la · LiveCloud · Eve Community
Follow us: @socialstrata
More information about the linux-fai