Hi,
Both subnets are in the same domain, but they each have a dhcp server( I
think they have to as dhcp is confined to a subnet (or?).
We made some progress on this by editing the dhcpd.conf on the 216
subnet, to look a little like the one on ranger. i.e. the following
lines:-
next-server 155.198.211.89;
option root-path "/ir62";
option option-151 "http://ranger";
The next-server line ensures that ranger(LCFG machine) disk is nfs
mounted rather than the disk on the 216 dhcp server.
LCFG fails for some other reason now, probably specific to the IBM setup
so we`ll wait for the cavalry to get back from the sysadmin camp.
Cheers,
Rod.
"Traylen, SM (Steve)" wrote:
>
> Barry,
>
> >As you may recall, we have two subnets in our group.
> >One is 155.198.211.x and one is 155.198.216.x.
> >The latter was introduced primarily for our IBM farm.
>
> >The LCFG server (ranger) is on the 211 subnet.
> >How do we get a node on the 216 subnet to get
> >a profile from ranger ? We've fiddled around a bit,
> >but we're finding it harder than expected.
> >Any ideas ?
>
> I am doing this myself at the moment and have almost
> been successful.
>
> Do the machines in the two subnets have the same domain,
> eg hep.ph.ic.ac.uk?
>
> Steve
>
> ----- Original Message -----
> From: "Traylen, SM (Steve) " <[log in to unmask]>
> To: "'Dr Barry MacEvoy'" <[log in to unmask]>
> Sent: Friday, April 19, 2002 2:02 PM
> Subject: RE: Jesus !
>
> > Okay,
> >
> > The intresting lines here are the first ones.
> >
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > lpr-0.50.1-1
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > pine-4.33-6.7
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > mutt-1.2.5i-8.6
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > root-3.02.04-1
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > glibc-2.1.3-22
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > aliroot-3.07.01-2
> > [WARNING] /usr/sbin/updaterpms: Couldn't find RPM header file for
> > sendmail-8.11.6-1.6.x
> >
> > The first of these says that you have requested lpr-0.50.1-1 but that
> > this rpm is not available in your repository.
> >
> > There are two reasons why what you have in the rpm lists and which rpms
> > are available could be different.
> >
> > i) New packages have been released and so you need to do a
> > createRepository.sh (*) to get the new ones.
> > You may as well do this , it only gets the new packages so should
> > not take very long at all.
> >
> > ii) A package is still in the list but has been removed from the Redhat
> > updates repository because there is a newer update.
> >
> > To solve this run createRepository to get the new rpms.
> >
> > If I now run
> >
> > $ find /opt/local/linux/6.2/RPM -name 'lpr-*' -print
> > /opt/local/linux/6.2/RPMS/release/lpr-0.50-4.i386.rpm
> > /opt/local/linux/6.2/RPMS/updates/lpr-0.50.5-1.i386.rpm
> >
> > At this point it is obvious that I am asking for lpr-0.50.1-1
> > but the newer lpr-0.50.5-1.i386.rpm is what is available.
> >
> > So you need to change what you are asking for by editing
> > /opt/local/linux/6.2/rpmcfg/rh6.2-updates.h problably
> > but it could be one of the other included files.
> >
> > This will fix the error caused by the differing lpr,
> >
> > Repeat the process for mutt, pine , root, ... glibc.
> >
> > The last of these is very significant and is what is causing all the
> > other errors about missing dependencies so they will go once glibc is
> > available for installation.
> >
> > I have to go through this with each release more or less but only for
> > the first node.
> >
> > You could create a /opt/local/linux/6.2/rpmcfg/my-extra-rpm.h
> > Put your additions in there and then include this file at the end of
> > /opt/local/linux/6.2/rpmcfg/UI-rpm.
> >
> > (*) There is now a replacement for createRepository ,
> > edg-updaterep-1.0-2.i386.rpm which you can get as rpm and
> > install on the lcfg box. It is on this page
> > http://www.lnl.infn.it/datagrid/wp4-install/
> > It is the same script, just made pretty , now has a manpage
> > the command is now '/usr/sbin/upaterep'
> >
> > Steve
> >
> >
> >
> > -----Original Message-----
> > From: Dr Barry MacEvoy [mailto:[log in to unmask]]
> > Sent: Friday, April 19, 2002 1:25 PM
> > To: Traylen, SM (Steve)
> > Subject: Jesus !
> >
> >
> > Hi Steve,
> >
> > All is not well here !
> >
> > When I run the make create, I get the attched list of messages. Any ideas
> ?!
> >
> > Thanks,
> >
> > Barry.
> >
> >
> >
> >
> > ----- Original Message -----
> > From: "Traylen, SM (Steve) " <[log in to unmask]>
> > To: "'Dr Barry MacEvoy '" <[log in to unmask]>; "Traylen, SM (Steve) "
> > <[log in to unmask]>
> > Sent: Thursday, April 18, 2002 9:02 PM
> > Subject: RE: FW: LCFG 'basic linux' install
> >
> >
> > >
> > > I forgot that bit, the default connection for cvs is rsh.
> > >
> > > export CVS_RSH=ssh
> > >
> > > and you should be fine, annoying it was the first thing though
> > > I expect.
> > >
> > > Steve
> > >
> > > -----Original Message-----
> > > From: Dr Barry MacEvoy
> > > To: Traylen, SM (Steve)
> > > Sent: 4/18/02 4:59 PM
> > > Subject: Re: FW: LCFG 'basic linux' install
> > >
> > > Hi Steve,
> > >
> > > Thanks for your prompt response.
> > >
> > > Although this sounds ridiculous, I now cannot seem to access the cvs
> > > repository !
> > >
> > > My /root/.ssh/config (as written by you) matches the one listed
> > > at:
> > >
> > > http://datagrid.in2p3.fr/HOWTO/CCIN2P3CVSaccessHOWTO-3.html#ss3.2
> > >
> > > but
> > >
> > > cvs -d datagrid.in2p3.fr:/cvs checkout -r v1_1_4 edg-release
> > >
> > > returns:
> > >
> > > ccsvrs02.in2p3.fr: Connection refused
> > > ccsvrs02.in2p3.fr: Connection refused
> > > trying normal rsh (/usr/bin/rsh)
> > > ccsvrs02.in2p3.fr: Connection refused
> > > cvs [checkout aborted]: end of file from server (consult above messages
> > > if
> > > any)
> > >
> > > Any ideas ?
> > >
> > > Thanks,
> > >
> > > Barry.
> > >
> > >
> > >
> > > ----- Original Message -----
> > > From: "Traylen, SM (Steve) " <[log in to unmask]>
> > > To: "'Dr Barry MacEvoy'" <[log in to unmask]>
> > > Sent: Thursday, April 18, 2002 3:12 PM
> > > Subject: RE: FW: LCFG 'basic linux' install
> > >
> > >
> > > >
> > > > You have all the rpms you need I think already.
> > > >
> > > > You need to checkout a copy of the edg-release (v1.1.4)
> > > >
> > > > I already added the bits you need to /root/.ssh/config
> > > >
> > > > go to /opt/local say.
> > > >
> > > > and
> > > >
> > > > $ cvs -d datagrid.in2p3.fr:/cvs checkout -r v1_1_4 edg-release
> > > >
> > > > This creates the directory edg-release which contains two
> > > > sub directories rpmcfg and source. The first has rpm lists for each
> > > machine
> > > > type, CE-rpm, SE-rpm, WN-rpm, ... The source directory contains the
> > > profile
> > > > for each machine type.
> > > >
> > > > Copy all the rpms lists for all the machine types into to the correct
> > > > location.
> > > >
> > > > $ cp edg-release/rpmlist/* /opt/local/linux/6.2/rpmcfg/.
> > > >
> > > > All the profile configuration files are in edg-release/source
> > > > and these need to placed in /var/obj/conf/profile/source/
> > > >
> > > > Take some care what you over right here. In particular do not
> > > > overwrite site-cfg.h.icl or djgpc(?). in this directory.
> > > > So that is
> > > > $ cp edg-release/source/* /var/obj/conf/profile/source/.
> > > >
> > > > maybe exclude updates-rpm.h since we had to fix that
> > > > to reflect newer updates that were available.
> > > >
> > > > Move to /var/obj/conf/profile/source/
> > > >
> > > > The CE example is testbed001 so copy that to
> > > > mynewce and edit mynewce.
> > > >
> > > > As a mimumum in mynewce you need to edit the HOSTNAME,
> > > > include your site-cfg.h.ic rather than the default one.
> > > > Also add a root password and ethernet card config rather like
> > > > was added to the end of the existing djgpc for the UI.
> > > >
> > > > The Next two things are what you always check before
> > > > attempting to install.
> > > >
> > > > Q1. Does the profile compile?
> > > > Q2. Is rpm list consistent?
> > > > ie do any rpms conflict and are they all available.
> > > >
> > > > A1. /usr/sbin/mkxprof -v -A mynewce
> > > >
> > > > A2. This is the make create bit in /opt/local/linux/6.2/rpmcfg
> > > >
> > > > Edit this make file so you are using CE-rpm as the top level
> > > > file.
> > > >
> > > > Run a make create and see what happens.
> > > >
> > > > Also before moving to a boot disk add the hardware address to the
> > > DHCP
> > > > server. To find the mac address I just boot with the floppy and then
> > > > check the errors in /var/log/messages on the DHCP server.
> > > >
> > > > It should be fairly smooth I would hope though.
> > > >
> > > > Steve
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > -----Original Message-----
> > > > From: Dr Barry MacEvoy [mailto:[log in to unmask]]
> > > > Sent: Thursday, April 18, 2002 2:40 PM
> > > > To: Traylen, SM (Steve)
> > > > Subject: Re: FW: LCFG 'basic linux' install
> > > >
> > > >
> > > > Hi Steve,
> > > >
> > > > I want to install a CE so that I can use Globus on my client.
> > > >
> > > > Could you please tell me the most painless way of doing this from
> > > where I
> > > > am at the moment ? I don't want to lose my ability to install a UI, so
> > > > presumably I need to download some more rpms and set new files
> > > > to generate the .xml...
> > > >
> > > > Cheers,
> > > >
> > > > Barry.
> > > >
> > > >
> > > >
> > > >
> > > > ---- Original Message -----
> > > > From: "Traylen, SM (Steve) " <[log in to unmask]>
> > > > To: "'Julian BLAKE'" <[log in to unmask]>; "Brian Coghlan"
> > > > <[log in to unmask]>
> > > > Cc: "Maite Barroso Lopez" <[log in to unmask]>; "German
> > > Cancio"
> > > > <[log in to unmask]>; "Traylen, SM (Steve) "
> > > > <[log in to unmask]>; "Emanuele Leonardi"
> > > <[log in to unmask]>;
> > > > "Barry MacEvoy (E-mail)" <[log in to unmask]>
> > > > Sent: Wednesday, April 17, 2002 10:41 AM
> > > > Subject: RE: FW: LCFG 'basic linux' install
> > > >
> > > >
> > > > > Julian,
> > > > >
> > > > > If you use the current 'client' config from
> > > > >
> > > > > config-defaults-rpmcfg and
> > > > > config-defaults-lcfg
> > > > >
> > > > > then what Brian describes where the machine hangs after the first
> > > > > reboot is exactly what happens.
> > > > >
> > > > > I confirmed this recently on a machine at Imperial College.
> > > > >
> > > > > If you afterwards mount the hard disk with a boot floppy to
> > > > > have a look what happened then the only thing on the
> > > > > hard disk is those files that have been created by the
> > > > > LCFG object scripts. There is not a single RPM that has been
> > > > > installed. I think this is the same as what you described
> > > > > was happening Brian. There is also nothing of any interest
> > > > > in /var/obj/log on the hard disk.
> > > > >
> > > > > The rpm config for client is consistent and will install so I guess
> > > > > there must be something weird in the client profile but I have
> > > never
> > > > > really looked to find the problem.
> > > > >
> > > > > Brian,
> > > > >
> > > > > If you go straight for a UI then the problem will hopefully
> > > disappear.
> > > > > So the current version is tag v1_1_4. This was what we did at
> > > > > Imperial and it worked a first time. I say a UI because its the
> > > > > quickest box.
> > > > >
> > > > > Steve
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > -----Original Message-----
> > > > > From: Julian BLAKE [mailto:[log in to unmask]]
> > > > > Sent: Wednesday, April 17, 2002 10:16 AM
> > > > > To: Brian Coghlan
> > > > > Cc: Maite Barroso Lopez; German Cancio; S M Traylen; Emanuele
> > > Leonardi
> > > > > Subject: Re: FW: LCFG 'basic linux' install
> > > > >
> > > > >
> > > > >
> > > > > Brian,
> > > > >
> > > > > I find your problem report confusing, as it seems to mix things
> > > which
> > > > > happen when you set up an LCFG server and things which happen when
> > > you
> > > try
> > > > > to use this server to install Linux on an LCFG client.
> > > > >
> > > > >
> > > > > Installing an LCFG server
> > > > > -------------------------
> > > > >
> > > > > The cookbook
> > > > >
> > > (http://datagrid.in2p3.fr/distribution/datagrid/wp4/installation/doc/
> > > > > -> recipe)
> > > > > is used by testbed 1 administrators at CERN to install Linux and
> > > LCFG on
> > > a
> > > > > new LCFG server computer. Having done this, they then complete the
> > > > > package repository on the server by adding all the higher level EDG
> > > > > software packages. Emanuele Leonardi has a script which mirrors
> > > software
> > > > > from datagrid.in2p3.fr.
> > > > >
> > > > > The purpose of section 8 of the cookbook is to install Red Hat 6.2
> > > updates
> > > > > packages, as well as OpenSSH and latex2html.
> > > > >
> > > > > If, at the time of running section 8, you have packages in
> > > > > /opt/local/linux/6.2/RPMS/external/ other than
> > > > >
> > > > > .latex2html-99.1-1.noarch.rpm
> > > > > .openssh-2.9p2-1.i386.rpm
> > > > > .openssh-clients-2.9p2-1.i386.rpm
> > > > > .openssh-server-2.9p2-1.i386.rpm
> > > > > Makefile
> > > > > latex2html-99.1-1.noarch.rpm
> > > > > openssh-2.9p2-1.i386.rpm
> > > > > openssh-clients-2.9p2-1.i386.rpm
> > > > > openssh-server-2.9p2-1.i386.rpm
> > > > >
> > > > > then you are not following the cookbook. If some of these other
> > > packages
> > > > > give you problems, it is likely that they were not designed to be
> > > used
> > > > > with Red Hat 6.2. A known case is version 2 of python. Red Hat 6.2
> > > uses
> > > > > python 1.5.2.
> > > > >
> > > > > The current version of the cookbook was last revised on 2nd April
> > > 2002.
> > > > > If your copy is older (revised in October 2001), you should get the
> > > new
> > > > > version.
> > > > >
> > > > > Having said all this, you can run Red Hat 6.2 release (instead of
> > > Red
> > > Hat
> > > > > 6.2 release plus updates) on your LCFG server. You won't have Red
> > > Hat's
> > > > > security bug fixes, but LCFG serving will work.
> > > > >
> > > > >
> > > > > LCFG client installation
> > > > > ------------------------
> > > > >
> > > > > If all you want to do is to install Linux and LCFG on an LCFG client
> > > > > computer, you can use the example package lists and LCFG resources
> > > from
> > > > > the config-defaults-rpmcfg and config-defaults-lcfg packages, which
> > > load
> > > > > files into /opt/local/linux/6.2/rpmcfg/ and
> > > /var/obj/conf/profile/source/
> > > > > respectively. In the first directory, the appropriate file is
> > > "client"
> > > > > In the second directory, you should first customize site.h, and then
> > > > > create a file "myclient" where myclient is the unqualified host name
> > > of
> > > > > your LCFG client computer, using the file testbed1 as example.
> > > > >
> > > > > For serious testbed 1 work, you should use package lists and LCFG
> > > > > resources from edg-release/ in CVS at IN2P3
> > > > > (http://datagrid.in2p3.fr/cgi-bin/cvsweb.cgi/edg-release/).
> > > > > See http://www.lnl.infn.it/~biasotto/datagrid/wp4-install/
> > > > > for useful documents on testbed 1 installation.
> > > > >
> > > > > Please refer to the WP4 LCFG FAQ
> > > > >
> > > (http://datagrid.in2p3.fr/distribution/datagrid/wp4/installation/doc/
> > > > > -> WP4 LCFG FAQ)
> > > > > for advice on setting up package lists and LCFG resources.
> > > > >
> > > > > It is likely that your update.rpmdir resource is incorrectly
> > > specified.
> > > > > The relevant code is in the subroutine UpdateRPMs() in
> > > obj-update/update
> > > > >
> > > (http://datagrid.in2p3.fr/cgi-bin/cvsweb.cgi/fabric_mgt/installation/
> > > > > obj-update/update). You may refer to this to find a match on the
> > > > > diagnostic message that you see. Another possibility is that your
> > > LCFG
> > > > > server is not exporting (via NFS) the package repository (see
> > > section 12
> > > > > of the cookbook).
> > > > >
> > > > > If LCFG installation fails in such a way that it brings up a shell,
> > > you
> > > > > can type
> > > > > qxprof update
> > > > > to see the values of resources of LCFG's update object, and
> > > > > qxprof update rpmdir
> > > > > to see the value of the update.rpmdir resource.
> > > > >
> > > > > ---
> > > > >
> > > > > Best wishes,
> > > > >
> > > > > Julian Blake.
> > > > >
> > > > > --
> > > > > Email: [log in to unmask], phone: +41 22 767 4893, fax: +41 22 767
> > > 4900.
> > > > > Post: CERN (European Organization for Nuclear Research), CH-1211
> > > Geneva
> > > > 23.
> > > > >
> > > >
> > > >
> > > ------------------------------------------------------------------------
> > > --
> > > > -
> > > > >
> > > > > On Tue, 16 Apr 2002, Maite Barroso Lopez wrote:
> > > > >
> > > > > > Hello Julian,
> > > > > >
> > > > > > Could you, please, have a look to this problem?
> > > > > > Thanks,
> > > > > >
> > > > > > Maite
> > > > > >
> > > > > > -----Original Message-----
> > > > > > From: Brian Coghlan [mailto:[log in to unmask]]
> > > > > > Sent: Monday, 08 April, 2002 11:35
> > > > > > To: Maite Barroso Lopez
> > > > > > Cc: [log in to unmask]; Traylen, SM (Steve)
> > > > > > Subject: LCFG 'basic linux' install
> > > > > >
> > > > > >
> > > > > > Maite, you might remember I had trouble with LCFG booting
> > > > > > a basic linux node (no edg software), which just hung after
> > > > > > the reboot after the rpm installs & commit, and promised
> > > > > > would investigate.
> > > > > > As far as I can see the major trouble is caused by running
> > > > > > make in the ./rpmcfg directory. This make has wildcards in
> > > > > > for rpms, so it installs the latest rpm that you have in
> > > > > > your release, updates and external directories - mostly
> > > > > > this runs into dependency problems - by the time you fix
> > > > > > them the whole thing hangs after rebooting after committing
> > > > > > the rpm updates. Despite a lot of time rebooting, I haven't
> > > > > > got any closer to understanding why.
> > > > > > It would be much better if that Makefile were changed to
> > > > > > install a particular consistent set of rpms.
> > > > > > However, I can't see why it's necessary to run the make
> > > > > > in the ./rpmcfg directory at all (stage 8 of Julian Blake's
> > > > > > install instructions) - the make in ./RPMS/LCFG installs
> > > > > > all the right rpms as far as I can see. So I would just
> > > > > > get rid of section 8 of Julian's document.
> > > > > > Without this there is much less a problem - the rpm installs
> > > > > > do not even start:
> > > > > > (a) If, say, I comment out the include for WorkerNode, it
> > > > > > just partitions the disk then reports it can't find the
> > > > > > file client_testbed, and falls back to the ramdisk
> > > > > > linux prompt.
> > > > > > (b) If, say, I comment out the includes for WorkerNode
> > > > > > and client_testbed-cfg-h, it just partitions the disk
> > > > > > then reports it can't find 'rpmdir', and falls back
> > > > > > to the ramdisk linux prompt.
> > > > > > (c) If, say, I comment out the includes for WorkerNode
> > > > > > client_testbed-cfg.h and the macros, it does the
> > > > > > same thing as (b), i.e. just partitions the disk then
> > > > > > reports it can't find 'rpmdir', and falls back to the
> > > > > > ramdisk linux prompt.
> > > > > > However, in none of these cases does it install the basic
> > > > > > linux configuration specified by EDG.
> > > > > > Regards
> > > > > > Brian
> > > > >
> > > >
> > >
> >
|
