July 2004 - openSUSE Autoinstall - openSUSE Mailing Lists

Still having trouble...
by Kevin Stallard 14 Jul '04

14 Jul '04

I have my autoinst.xml set up as follows <software> <base>swos</base> </software> and swos.sel is in my selections file and directory.yast file in <CDROOT>/suse/setup/descr For some reason, I get a "cannot set base package selection 'swos'" error when I try to install from this CD. The swos.sel file looks like this... # SuSE-Linux-Package-Selection 3.0 -- (c) 2002 SuSE Linux AG # generated on Wed Apr 7 13:41:25 UTC 2004 =Ver: 3.0 =Sel: swos =Sum: Base swos System =Cat: base =Vis: true =Ord: 01 What am I missing? I can't find anything in the documentation that indicates what I'm doing wrong. -K -- Kevin Stallard <kstallard(a)lancope.com>

2 1

chroot scripts using <location>nfs:// fail
by Frank Steiner 14 Jul '04

14 Jul '04

Hi, chroot-Scripts can't be specified using the <location>nfs://... tag. When AY mounts the given NFS location to copy the chroot scripts, it mounts the directory on /mnt, thus hiding the already installed system. Since AY tries to copy the chroot scripts *into* the installed system, i.e., to /mnt/var/adm/..., this must fail or course. You can see this problem in y2log: 2004-07-14 14:42:12 <1> linux(1121) [agent-system] SystemAgent.cc(Execute):1094 mkdir /mnt/var/adm/autoinstall/scripts 2004-07-14 14:42:12 <1> linux(1121) [agent-system] SystemAgent.cc(Execute):1094 mkdir /mnt/var/adm/autoinstall/logs 2004-07-14 14:42:12 <1> linux(1121) [YCP] AutoinstScripts.ycp:375 Writing chroot script into /mnt/var/adm/autoinstall/scripts/mychroot.script 2004-07-14 14:42:13 <3> linux(1121) [bash] ShellCommand.cc(shellcommand):73 /bin/cp: cannot create regular file `/mnt/var/adm/autoinstall/scripts/mychroot.script': No such file or directory That's the point where the target path /mnt/var/adm/autoinstall/scripts/ has been hidden by mounting the NFS directory. Since we can fetch stuff only via NFS (no http or whatsoever), we can't use locations for chroot scripts at all at the moment. Is there a change that this could be patched for the current release (not just for the next release)? I would also be happy about an inofficial patch! cu, Frank -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

1 0

DHCP Fails only on second install on same hardware
by Nicholas DeClario 14 Jul '04

14 Jul '04

Hey there, I am doing multiple installs on the same hardware via autoyast as I try different configurations. My setup is boot from a PXE server, retrieve my autoyast control file via http and retrieve my SuSE packages via http. My DHCP server is also my tftp server hosting my PXE setup while the http server, hosting the autoyast control file and SuSE packages are on a different server. I run the system through an install, everything's fine. Then, on a second try the system will get it's DHCP information to boot from the PXE server, the kernel and initrd will start up and then attempt to get the DHCP information from within linux at this point where it hangs; however, if I jump to a different tty to see the logs I can see it did indeed get a valid interface configuration from the DHCP server and I can further confirm this by checking the logs; unfortunatly, at this point it does not contact my httpd server for the control file or the packages. I tail the logs on that server and they come up empty. Eventually it will time out and give me an error saying auto installation is impossible and force me into a manual install. Now, if I attempt to do this exact same thing on a new piece of hardware it works fine, until the second time I try to go through an install. Then I can reproduce the problem exactly. Is something changing on the system that autoyast is picking up after an initial install that is preventing a second install from happening? Is there perhaps a problem with my DHCP server? Any help would be much appreciated. Thanks, -Nick

1 2

disk parameter in lilo.conf
by Frank Steiner 14 Jul '04

14 Jul '04

Hi, I tried to include the lines disk=/dev/sda bios=0x80 disk=/dev/hda bios=0x81 into lilo.conf. I found it working this way: <bootloader> <global config:type="list"> <global_entry> <key>disk</key> <value>disk=/dev/sda bios=0x80</value> </global_entry> <global_entry> <key>disk</key> <value>disk=/dev/hda bios=0x81</value> </global_entry> </global> ... </bootloader> However, from the description on the web page I was afraid this would end up as disk = "disk=/dev/sda bios=0x80" like it does with the other key/value pairs which always come out as key = "value". Of course, it is great that it works this way! I just wanted to ask if this is intended (since I found no hint in the docu) or just a bug. If it is a bug, it should be recognized a feature and kept in future versions :-)) The way it works now allows us to install everything on our servers that have scsi and ide disks, which is great since it did not work in SuSE 9.0 :-))) cu, Frank -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

1 0

Variables in result part in rules
by Frank Steiner 14 Jul '04

14 Jul '04

Hi, I played around with using variables like @bla@ in the result part of a rule. I tried to do sth. like this: 1) whatever k_arch we match, use @k_arch@.xml as profile. I could not do that because I could not reach that "whatever we match". When I just remove the <match> element, AY seems to take this as "undefined" and tries to match "undefined" against the value of the variable k_arch, which fails. However, if I need to match explicitely against e.g. "kernel-default", then I don't need the @k_arch@ variable anymore. This would only make sense if I don't now what exactly I matched against. This would work only when using a custom script to detect the hostip and then use a "range" match type. But for strings like k_arch I don't know how to achieve this. One would need sth. like "<match>*</match>" to match against anything, or AY should take a missing match as "always succeed". Or is there any other way to achieve this? 2) I tried to use a profile <hostname>.xml for whatever hostname we matched. Since hostname contains the IP at the moment, I tried a custom script like this: <rule> <custom1> <script> <![CDATA[ #!/bin/sh export myhostname=`grep HOSTNAME /var/lib/dhcpcd/dhcpcd-eth0.info | sed 's/HOSTNAME=//g'` echo -n undefined ]]> </script> <match>undefined</match> <match_type>exact</match_type> </custom1> <result> <profile>@myhostname@.xml</profile> <continue config:type="boolean">true</continue> </result> </rule> From y2log I see that the matching succeeds, and "myhostname" is stated as variable (y2log states "var: myhostname"), but it's value is empty, resulting in a load for ".xml". Is there a way to transfer variables from a custom-script into the environment? Otherwise, again, it seems that using variables in the result element makes indeed only sense for ranges... cu, Frank -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

1 0

hostname contains IP?
by Frank Steiner 14 Jul '04

14 Jul '04

Hi, I'm trying to match against the hostname in rules.xml: <rule> <hostname> <match>zassenhaus</match> <match_type>exact</match_type> </hostname> <result> <profile>profiles/test.xml</profile> <continue config:type="boolean">true</continue> </result> </rule> But it does not work. When I look into y2log I find this line: 2004-07-14 08:50:43 <1> linux(1121) [YCP] AutoInstallRules.ycp:374 Bash return: if [ "$hostname" = "zassenhaus" ]; then exit 0; else exit 1; fi ($["exit":1, "stderr":"", "stdout":""]) ($["arch":"i386", "hostaddress":"141.84.1.172", "hostname":"141.84.1.172", "karch":"kernel-default"]) So it looks like that "hostname" has the value ".141.84.1.172" in the parameter list while I would expect it to have the hostname zassenhaus. dhcp returns the hostname, I verified that with tcpdump. Is this a bug in AY or do I have to tweak sth. special in the dhcp server? cu, Frank -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

1 0

Mount timeout
by J. David Eisenberg 14 Jul '04

14 Jul '04

I am still trying to get an NFS install to work. My info file says: Install: nfs://205.155.157.16/usr/local/SuSE AutoYaST: http://205.155.157.16/profiles/autoinst.xml Serverdir: /usr/local/SuSE Server: 205.155.157.16 DHCP: 1 UseDHCP: 1 Netdevice: eth0 Netmask: 255.255.255.0 Gateway: 205.155.157.254 Nameserver: 198.189.12.23 InstMode: nfs Textmode: 0 and the y2log says this: ==== 2004-07-13 10:37:36 <1> linux(968) [wfm++] ycpTools.cc(load):76 SourceCreate( YT_STRING[, YT_STRING] ): ["nfs://205.155.157.16/usr/local/SuSE", ""] 2004-07-13 10:37:36 <1> linux(968) [packagemanager] InstSrcManager.cc(scanMedia):412 scanMedia nfs://205.155.157.16/usr/local/SuSE 2004-07-13 10:37:36 <1> linux(968) [mediaaccess] MediaAccess.cc(open):114 Opened: nfs(nfs://205.155.157.16/usr/local/SuSE not attached; localRoot "/var/adm/YaST/InstSrcManager/tmp") 2004-07-13 10:37:56 <1> linux(968) [mediaaccess++] Mount.cc(mount):98 stdout: mount: RPC: Timed out 2004-07-13 10:37:56 <2> linux(968) [mediaaccess] Mount.cc(mount):130 mount 205.155.157.16:/usr/local/SuSE /var/adm/YaST/InstSrcManager/tmp: ERROR(Media:mount failed) 2004-07-13 10:37:56 <2> linux(968) [mediaaccess] MediaHandler.cc(attach):165 Attach failed: ERROR(Media:mount failed) nfs://205.155.157.16/usr/local/SuSE not attached; localRoot "/var/adm/YaST/InstSrcManager/tmp" 2004-07-13 10:37:56 <3> linux(968) [packagemanager] InstSrcManager.cc(scanMedia):424 Failed to attach media: ERROR(Media:mount failed) ==== However, if I go to ALT-F2 and type the command mount 205.155.157.16:/usr/local/SuSE /var/adm/YaST/InstSrcManager/tmp the command succeeds. I don't have a firewall active on my machine at 205.155.157.16. What am I doing wrong? -- J. David Eisenberg http://catcode.com/

2 1

Partitioning with existing fstab
by Frank Steiner 13 Jul '04

13 Jul '04

Hi Anas, first of all: in the dtd-files (from autoyast2-installation.rpm) there are no definitions for all the "partitioning_advanced" stuff are missing. About the feature itself I have one question and one proposal: - How can one implement the following scenario: If there is a fstab, use it. If no fstab is found, go on with some other profiles, i.e. format according to the size of the hard disk. I'm not sure how to define this in the profile. - I would vote for a change in the default behaviour of formatting! AY1 used to leave all found partitions untouched and just format those, that were explicitely specified. Now with AY2 doing this the other way around, the feature is very dangerous: I have about 60 machines here, most of them having the same hard disk partitions. But there are always a handfull of machines that have Raids attached, additional hard disks etc. What I want to do is leaving all partitions untouched,but just format "/" and "/boot" on every machine. For that, I would need only one profile for all hosts, if the partitions were not formatted automatically. The current behaviour of AY2 is dangerous in that I might once forget to specify a partition that exists on some hosts, and then it will eventually get formatted. So, maybe a boolean flag in the <fstab> element, specifying if formatting or not-formatting is the default, could help? cu, Frank -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

1 0

Re: Antwort: [suse-autoinstall] AY on SuSE 9.1: only textmode
by Frank Steiner 13 Jul '04

13 Jul '04

Joachim.Tork(a)gad.de wrote: > Hi, > > try > > append initrd=initrd textmode=1 Hm, no, you got me wrong: I'm not trying to run in textmode. I'm trying to run yast2 in graphical mode, but it always starts in textmode automatically telling me that the X server was already active (which it is not, of course :-() -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

3 4

Bug in runlevels.dtd
by Frank Steiner 13 Jul '04

13 Jul '04

Hi Anas, in the runlevels.dtd for SuSE 9.1 there is a line <!ELEMENT default (#PCDATA)> missing. cu, Frank -- Dipl.-Inform. Frank Steiner Web: http://www.bio.ifi.lmu.de/~steiner/ Lehrstuhl f. Bioinformatik Mail: http://www.bio.ifi.lmu.de/~steiner/m/ LMU, Amalienstr. 17 Phone: +49 89 2180-4049 80333 Muenchen, Germany Fax: +49 89 2180-99-4049 * Rekursion kann man erst verstehen, wenn man Rekursion verstanden hat. *

1 0