Hi folks. Well, I managed to get 3.6.3 working. 3.7.6 does not.
Hopefully the information I give here in this (lengthly-sorry) post will
help the developers figure out what's going on..... Hardware = HP
rx2600 Integrity (IA64). OS= RedHat AS 4.
First, a note: In my latest uses of systemimager (this IA64 cluster, but
also others) it seems to me that any of the distributed RPM's have
issues. I have always seemed to have to build my own RPM's to solve a
multitude of problems. Not sure why, but if anyone out there is having
problems, that is the first place to start.......
OK, so I grabbed 3.7.6, built it, and was having a devil of a time
getting my scsi (fusion MPT) disk controller to be properly recognized.
The main problem I was having was the (somewhat popular) "cannot stat
/dev/sda" problem. While I _initially_ thought that the fusion MPT
scsi drivers weren't getting loaded-- it actually appears that they
ARE! Here's the boot console output:
Loading cmd64x...
CMD649: 100% native mode on irq 57
ide0: BM-DMA at 0x0d40-0x0d47, BIOS settings: hda:pio, hdb:pio
ide1: BM-DMA at 0x0d48-0x0d4f, BIOS settings: hdc:pio, hdd:pio
hda: DV-28E-C, ATAPI CD/DVD-ROM drive
ide0 at 0xd58-0xd5f,0xd66 on irq 57
hda: ATAPI 24X DVD-ROM drive, 256kB Cache, UDMA(33)
Uniform CD-ROM driver Revision: 3.20
Fusion MPT base driver 3.03.07
Copyright (c) 1999-2005 LSI Logic Corporation
done!
Loading mptscsih...
done! <==== here you can see that mptscsih DOES
get loaded after all!
Loading mptscsih...
done!
Loading ohci-hcd...
not needed or already loaded.
Loading ohci-hcd...
not needed or already loaded.
Loading ehci-hcd...
not needed or already loaded.
get_hostname_by_hosts_file
Hosts file exists...
Searching for this machine's hostname in /scripts/hosts by IP:
192.168.96.103
This hosts name is: wolf03
run_pre_install_scripts
>>> 99all.harmless_example_script
I live in /var/lib/systemimager/scripts/pre-install.
choose_autoinstall_script
Using autoinstall script: /scripts/wolf03.sh
write_variables
run_autoinstall_script
>>> /scripts/wolf03.sh
get_arch
Partitioning /dev/sda...
Old partition table for /dev/sda:
Error: Could not stat device /dev/sda - No such file or directory.
<<<==== BUT, still no disk device. ??????
dd if=/dev/zero of=/dev/sda bs=512 count=1 || shellout
1+0 records in
1+0 records out
blockdev --rereadpt /dev/sda
BLKRRPART: Inappropriate ioctl for device
parted -s -- /dev/sda mklabel gpt || shellout
Error: Can't have the end before the start!
Error: Can't have a partition outside the disk!
Error: Invalid argument during seek for write on /dev/sda
Error: Can't have a partition outside the disk!
Killing off running processes.
write_variables
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>
I also tried the following:
* Adding the INSMOD commands into the proper location - which ended up
simply causing the system to HANG when loading MPTSCSIH. I sent an
email last week with that problerm. No idea why, and no error messages
to go on here.
* Tried UYOK- with the current kernel from the running golden client--
multiple problems here, as the system couldn't seem to locate its
modules correctly. Documentation on this is thin, so I am not sure
what was going on.
SO.... I reverted back to 3.6.3 (again, built it all myself), and
things work pretty much out of the box (!!!!!!).
Console output is as follows (and, we're good- and running). MORE
IMPORTANT INFO FOLLOWS CONSOLE TEXT
Loading sd_mod...
not needed or already loaded.
Loading ide-disk...
not needed or already loaded.
Loading e100...
not needed or already loaded.
Loading tg3...
not needed or already loaded.
Loading cmd64x...
not needed or already loaded.
Loading mptscsih...
not needed or already loaded.
Loading mptscsih...
not needed or already loaded.
Loading ide-scsi...
done!
Loading usb-ohci...
not needed or already loaded.
Loading usb-ohci...
not needed or already loaded.
Loading ehci-hcd...
not needed or already loaded.
get_hostname_by_hosts_file
Hosts file exists...
Searching for this machine's hostname in /scripts/hosts by IP:
192.168.96.103
This hosts name is: wolf03
run_pre_install_scripts
>>> 99all.harmless_example_script
I live in /var/lib/systemimager/scripts/pre-install.
choose_autoinstall_script
Using autoinstall script: /scripts/wolf03.sh
write_variables
run_autoinstall_script
>>> /scripts/wolf03.sh
get_arch
DISKORDER=sd,cciss,ida,rd,hd
enumerate_disks
sda
sdb
hda
DISKS=3
Partitioning /dev/sda...
Old partition table for /dev/sda:
program parted is using a deprecated SCSI ioctl, please convert it to
SG_IO
Disk geometry for /dev/scsi/host0/bus0/target0/lun0/disc: 0kB - 73GB
Disk label type: gpt
Number Start End Size File system Name Flags
1 17kB 105MB 105MB fat16 boot
2 105MB 524MB 419MB
hp-service
3 524MB 71GB 71GB ext3
4 71GB 73GB 2089MB linux-swap
parted -s -- /dev/sda mklabel gpt || shellout
program parted is using a deprecated SCSI ioctl, please convert it to
SG_IO
SCSI device sda: 143374738 512-byte hdwr sectors (73408 MB)
SCSI device sda: drive cache: write through
/dev/scsi/host0/bus0/target0/lun0:
program parted is using a deprecated SCSI ioctl, please convert it to
SG_IO
Creating partition /dev/sda1.
parted -s -- /deprogram parted is using a deprecated SCSI ioctl, please
convert it to SG_IO
v/sda mkpart primary ext2 0 99.864 || shellout
SCSI device sda: 143374738 512-byte hdwr sectors (73408 MB)
SCSI device sda: drive cache: write through
/dev/scsi/host0/bus0/target0/lun0: p1
parted -s -- /dev/sda set 1 boot on
program parted is using a deprecated SCSI ioctl, please convert it to
SG_IO
SCSI device sda: 143374738 512-byte hdwr sectors (73408 MB)
SCSI device sda: drive cache: write through
/dev/scsi/host0/bus0/target0/lun0: p1
Creating partition /dev/sda2.
parted -s -- /dev/sda mkpart priprogram parted is using a deprecated
SCSI ioctl, please convert it to SG_IO
mary ext2 99.864 4195.864 || shellout
SCSI device sda: 143374738 512-byte hdwr sectors (73408 MB)
SCSI device sda: drive cache: write through
...and so on. Voila. The final problem I ran into was the (again,
popular) problem of tg3 and e100 getting loaded BACKWARDS. I believe
this has something to do with how the pci bus is probed. Anyway, I
went looking for information on systemconfigurator, but I have no
/etc/systemconfig directory and was not clear if the .conf file(s)
belonged on the client, or server-- and whether or not
systemconfigurator even used these files for this? Only reference to
the systemconfigurator config file I could find was in this forum 3
years ago. I instead chose a slightly simpler solution, which was to
add apost-install script that really does one thing: copies a saved
"modprobe.conf.save" to "modprobe.conf". - which has the lovely side
effect of basically IGNORING whatever systemimager does to my modprobe,
and puts the original back. Voila-- tg3 is now (properly) associated
with eth0, and e100 with eth1. I don't exactly understand why
systemimager gets this backwards.
OK, so summarized:
* I can not get 3.7.6 to boot an IA64 client- period. Any insight
would be great- I'd be most willing to test stuff....
* 3.6.3 for the IA64 systems seems just fine.
* UYOK stuff could use some additional documentation... :-)
* I used a post-install script to solve the annoying tgg/e100 eth0/eth1
ordering problem.....
---------------------------------------------------------
Kent C. Brodie - [EMAIL PROTECTED]
Department of Physiology
Medical College of Wisconsin
(414) 456-8590
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users
