Anyone know what's up w/ trying to install bind through kickstart? I am trying all variations of group installs, individual package installs and it just won't install, it either fails on the libs, the chroot pkg or whatever?
On Tue, 2010-01-05 at 04:01 +0000, Joseph L. Casale wrote:
Anyone know what's up w/ trying to install bind through kickstart? I am trying all variations of group installs, individual package installs and it just won't install, it either fails on the libs, the chroot pkg or whatever? _______________________________________________
--- OK using KS.. But from cd disks or http or ftp or nfs? Make sure you do a self dependancy check on bind itself. Then I would check for the correct gpg key on bind it self and deps. An iffy cdrom or network connections that are some times faulty may cause it. Usually with me turns out to be a semi bad cdrom or corrupt package.
You can switch through to the anaconda terminal that shows all the errors and that would be a good start to check first hand.
John
OK using KS.. But from cd disks or http or ftp or nfs? Make sure you do a self dependancy check on bind itself. Then I would check for the correct gpg key on bind it self and deps. An iffy cdrom or network connections that are some times faulty may cause it. Usually with me turns out to be a semi bad cdrom or corrupt package.
You can switch through to the anaconda terminal that shows all the errors and that would be a good start to check first hand.
I was testing it an esxi server (can't change consoles:<) from my wkst and performing the install over vnc.
It only failed when using the dvd or iso to start a cdrom install with the updates repo supplied. I'll retest today, must be an issue with telus repo...
Install works when type set to url though?
iso and dvd check out fine.
OK using KS.. But from cd disks or http or ftp or nfs? Make sure you do a self dependancy check on bind itself. Then I would check for the correct gpg key on bind it self and deps. An iffy cdrom or network connections that are some times faulty may cause it. Usually with me turns out to be a semi bad cdrom or corrupt package.
You can switch through to the anaconda terminal that shows all the errors and that would be a good start to check first hand.
Ok, finally got on a windows wkst so I could vm this up and be able to switch consoles.
Looks like the problem is related to using two nics. Even though I specify ksdevice=eth0 on the grub line, it still seems to have trouble for some parts of downloading packages but not all?
On one of the consoles, I see a dhclient help screen? I guess Anaconda is doing something with incorrect syntax.
My ks lines are: network --device=eth0 --bootproto=dhcp network --device=eth1 --bootproto=static --ip=192.168.13.1 --netmask=255.255.255.0 --hostname=foo.bar.com
So whats wrong here, or how can I resolve this? This *only* manifests when using cdrom and repo, but if I do a url install, it works?
jlc
Ok, finally got on a windows wkst so I could vm this up and be able to switch consoles.
I should have paid more attention, I spoke to soon. Problem is elsewhere...
It failed again, so, from the console during install while Anaconda has errored out, I type `wget http://something_on_my_server and it downloads the file.
If I type `http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4...' it downloads it but it doesn't appear anywhere? If I type `http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4... -O foo'
I get a foo in /? WTF is this all about?
On Wed, 2010-01-06 at 23:17 +0000, Joseph L. Casale wrote:
Ok, finally got on a windows wkst so I could vm this up and be able to switch consoles.
I should have paid more attention, I spoke to soon. Problem is elsewhere...
It failed again, so, from the console during install while Anaconda has errored out, I type `wget http://something_on_my_server and it downloads the file.
If I type `http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4...' it downloads it but it doesn't appear anywhere? If I type `http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4... -O foo'
I get a foo in /? WTF is this all about?
---
Maybe your networking somewhere. Replace the rpm or reburn the disk. A disk can check good and still be bad!
[ebla@bla Desktop]$ wget http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4... --2010-01-06 18:38:41-- http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4... Resolving www.muug.mb.ca... 130.179.31.46 Connecting to www.muug.mb.ca|130.179.31.46|:80... connected. HTTP request sent, awaiting response... 200 OK Length: 45017 (44K) [application/x-rpm] Saving to: `bind-chroot-9.3.6-4.P1.el5_4.1.x86_64.rpm'
100%[===================================================================================================>] 45,017 9.38K/s in 4.8s
2010-01-06 18:38:47 (9.24 KB/s) - `bind-chroot-9.3.6-4.P1.el5_4.1.x86_64.rpm' saved [45017/45017]
John
Maybe your networking somewhere. Replace the rpm or reburn the disk. A disk can check good and still be bad!
I don't follow you:)
This is a kickstart install from a cdrom with a url line pointed to a repo. Its going across the wire for the bind rpm, that's why its using wget...
I did not only verify the hash, but used another iso and tried a physical machine with a burned dvd. Same behavior.
There is not a network issue, see the multiple wget commands and the results. - wget for a file on some other server works. - wget for the rpm appears to work, but the actual file is no where - wget for that *same* rpm, but with a -O foo works as well, but I now have a "foo" file to show for it:)
On Wed, 2010-01-06 at 23:53 +0000, Joseph L. Casale wrote:
Maybe your networking somewhere. Replace the rpm or reburn the disk. A disk can check good and still be bad!
I don't follow you:)
This is a kickstart install from a cdrom with a url line pointed to a repo. Its going across the wire for the bind rpm, that's why its using wget...
I did not only verify the hash, but used another iso and tried a physical machine with a burned dvd. Same behavior.
There is not a network issue, see the multiple wget commands and the results.
- wget for a file on some other server works.
- wget for the rpm appears to work, but the actual file is no where
- wget for that *same* rpm, but with a -O foo works as well, but I now have a "foo"
file to show for it:)
--- Have you tried another install tree? for bind ... Only other thing I can think of is bad NIC enumeration sequence. older kernels have that problem, dhcp dns issues? does it fetch other rpms fine with wget? any way to use tcp dump for network trouble shooting?
John
On Wed, 2010-01-06 at 23:17 +0000, Joseph L. Casale wrote:
Ok, finally got on a windows wkst so I could vm this up and be able to switch consoles.
I should have paid more attention, I spoke to soon. Problem is elsewhere...
It failed again, so, from the console during install while Anaconda has errored out, I type `wget http://something_on_my_server and it downloads the file.
If I type `http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4...' it downloads it but it doesn't appear anywhere? If I type `http://www.muug.mb.ca/pub/centos/5.4/updates/x86_64/RPMS/bind-chroot-9.3.6-4... -O foo'
I get a foo in /? WTF is this all about?
--- Another thought. I just laughed at myself. Yours went into outer space! There is no saving spot for it?
John
-
JohnS -
Joseph L. Casale