From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8+) Gecko/20020207 Description of problem: Using netboot.img I mounted a directory containing the 3 binary ISO images. anaconda immediately fails with the following messages: need to look at ['beta1-i386-disc1.iso', 'beta1-i386-disc3.iso', 'beta1-i386-disc3.iso'] found {1: 'beta1-i386-disc1.iso', 2: 'beta1-i386-disc2.iso', 3: 'beta1-i386-disc3.iso'} in tree mnt/source Traceback (most recent call last): File "/usr/bin/anaconda", line 608 in ? iutil.makeDriveDeviceNodes() File "/usr/lib/anaconda/iutil.py", line 417, in makeDriveDeviceNodes isys.makeDevInode(drive,"/dev/%s" % (drive,)) File "/usr/lib/anaconda/isys.py", line 237 in makeDevInode _isys.mkdevinode(name,fn) SystemError: (2, 'No such file or directory') Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1.boot with boonet.img boot disk 2.provide info for NFS directory containing the 3 ISOs 3. Actual Results: the error message above Expected Results: installer continuing Additional info:
Tried to use the boot.img and use CDROM images with the same result. So I gues it's the hardware combination. Which is as follows: - the machine has 3 SCSI CDROMs, *no* IDE CDROM - it has an Adaptec 29xx controller - I cannot boot from CDROM due to the lack of IDE CROMs - I used bootnet.img and boot.img, both fail with the same error Without a fix for this I cannot do anything. It's really important that I can so please make a solution available ASAP.
Forgot to mention: - the two harddrives in the machines are IDE (yeah, I know, strange machine...)
More information: if I select the text installer it works.
Was there anything else after the last line of the error message you gave above? Usually there is a dump of all the python interpretter state which would be helpful, especially if you can safe to floppy. You could try disconnecting the SCSI cdrom drives if you really need to install it asap.
What I wrote down was everything there was to see. And I got the beta installed by using the TUI installer.
Is this any better with beta 2?
With beta2 the results are not much better. Instead of an anaconda message I get a kernel message. To the right of the text line in the box (this is TUI) Kernel paic: Loop 1 and the machine is frozen. After lots of SysRq key pressing I'm not sure I caught the right process but what I saw is this eip 0010:[c0116b8c] eax: 00000000 ebx e0c7db31 ecx c022668f edx 00000000 esi 00000000 edi 00000000 ebp 00000000 ds 0018 es 0018 cr0 8005003b cr2 40003090 cr3 1f8a8000 cr4 000002d0 Call Trace: e0c7478a e0c7db31 e0c7db14 e0c69f26 e0c6a4a7 e0c12498 e0c129db e0c1303e c01086bd e0c70018 c0106e72 e0c12f60 I copies this by hand so there might be bugs in there. In case you want to know, the server is a RH7.2 box Oh, and I think I haven't mentioned this is a dual PII Xeon box.
Well, I tried it again. Now it worked fine. The difference to the first attempt was that first I mistyped the IP address of the machine which has is going to be installed. Trying to mount the NFS directory failed in this case since there was no route to the host. After going back and correcting the IP address the mounting succeeded but the above mentioned kernel panic stopped everything. During the second attempt I entered the correct IP address right away, didn't have failed NFS mount attempts, and everything is fine.
Arjan, any thoughts on the oops?
Ulrich, are you still seeing this issue with the invalid IP address with Beta 3 (Skipjack)? I'm not able to replicate here in the office, but didn't try with a 'no route to host address' just a 'destination not reachable'
I tried to provoke the error on the same machine with, what I think, are the same steps I did before. It worked fine, no backtrace. This is with beta4. My only issue is that the error message isn't the best. The kernel has more information. Maybe it could be improved.
There's no way we can change the string now (string freeze is long gone) and figuring out deeper causes of obscure error messages hit such a small percentage of the time isn't really worth the likelihood of breaking (since we can't reproduce it, we can't test that what we do really works)