Netboot Mailing List (by thread)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: etherboot & NFS problem - part of the problem solved




On Fri, Nov 03, 2000 at 10:03:53AM +0100, Christoph Plattner wrote:

>
>Hi !
>
>No, No. Memory leak on the client of course !
>
>But, more ....
>So from the "distance" I have no idea.
>
>Perhaps it is a good idea to find anywhere GRUB (netbooted or booted via
>disk).
>It is really a good chance to "play around", otherwise you such long
>turn around
>cycles....

Anyway Chris. Thanks a million for all the help. I will see what I can do.

Warm Regards


Rajesh

>
>Rajesh Fowkar wrote:
>> 
>> Hi Chris,
>> 
>> Thanks for sparing so much time to help me solve this etherboot problem.
>> 
>> >
>> >I think, this looks like a memory leak. How much memory do you have ?
>> >Check if you have added a "mem=" kernel parameter by accident.
>> 
>> The Linux server has got 256MB RAM. It is a PIII 550MHZ machine with RAID1 configured(Mirroring). Not added any mem= kernel parameters because without doing that the memory is detected as 256MB.
>> 
>> -----------------------------------------------------------------------------------
>> [root@rhlinux rajeshpc]# free
>>              total       used       free     shared    buffers     cached
>> Mem:        257680     198952      58728      61216      37040     118864
>> -/+ buffers/cache:      43048     214632
>> Swap:       512888          0     512888
>> [root@rhlinux rajeshpc]#
>> -----------------------------------------------------------------------------------
>> 
>> >What tool have you used to prepare the kernel. Wich parameter du you
>> >have used ?
>> 
>> make xconfig.
>> Thinking that I might have done something wrong while compiling the kernel, I recompiled using the following options : ( I have indicated the options selected in brackets )
>> 
>> -----------------------------------------------------------------------------------
>> -> Processor type & features
>>         - Processor Family (386)
>>         - Disable PIII Serial No. (Y)
>>         - Maximum Physical Memory (1GB)
>> 
>> -> Block Devices
>>         - Disabled options for RAID
>> 
>> -> Networking options
>>         - IP Kernel level autoconfiguration
>>                 - BOOTP Support (Y)
>>                 - RARP Support (Y)
>> -> Ethernet ( 10 or 100 Mbit )
>>         - NE2000/NE1000 Support (Y)
>> 
>> -> Appletalk Devices
>>         - Support (N)
>> 
>> -> Joystick Support (N)
>> 
>> -> Network File System
>>         - NFS Filesystem Support (Y)
>>         - Root Filesystem on NFS (Y)
>> 
>> -> Sound
>>         - Sound Card Support (N)
>> -----------------------------------------------------------------------------------
>> 
>> ->
>> >The interesting points are:
>> >
>> >       "VM: killing process XXXXX"
>> 
>> These errors are still repeating.
>> 
>> Than it gives the following :
>> 
>> Mounting /proc                  [Failed]
>> Starting Kernel Parameters      [Failed]
>> 
>> >Perhaps it is a good idea to check the kernel itself. For example, if
>> >your
>> >target (client) has a floppy drive, do following hack. Do a `dd' of the
>> >kernel
>> >on a floppy
>> >       dd if=/tftpboot/.../vmlinuz of=/dev/fd0 bs=512
>> >and boot from floppy. The create a "dummy" device, which has the major 0
>> >and
>> >minor 255, like
>> >       mknod /dev/nfs c 0 255
>> >on the linux host. After that use
>> >       rdev /dev/fd0 /dev/nfs
>> >... so the kernel is patched to boot from major 0 and minor 255.
>> >(Another way is to edit with a hex editor the space, where the kernel
>> >saves
>> >major and minor of the boot device).
>> >
>> >Boot the client from floppy and everything should work (?!)
>> 
>> I have done all this. But when I try to boot the node from this floppy nothing happens. Just floppy is accessed and than nothing happens.
>> 
>> >AND: Before you do anything, read the kernel informations on the screen,
>> >when
>> >ther kernel starts to boot, if there is anything, which seems to be a
>> >problem.
>> 
>> Only problem which I found on the node was it tries to access hda which it cannot find and than it gies hda disk error and than the NFS-root mounting starts. No other error.
>> 
>> Thanks
>> 
>> Warm Regards
>> 
>> Rajesh
>> 
>> >Rajesh Fowkar wrote:
>> >>
>> >> Hi,
>> >>
>> >> I tried everything. Copying /etc/ld.so.cache to /tftpboot/rajeshpc/etc.
>> >>
>> >> Used ldconfig -r /tftpboot/rajeshpc
>> >>
>> >> But still I am getting the following messages on the Diskless node :
>> >>
>> >> -----------------------------------------------------------------------------
>> >> VFS: Mounted root(NFS filesystem)
>> >> Freeing unused kernel memory : 76K freed
>> >> INIT: version 2.78 booting
>> >> VM: killing process rc.sysinit
>> >> INIT: Entering runlevel: 3
>> >> bash: out of virtual memory!
>> >> VM: killing process rc
>> >> INIT: no more processes left in this runlevel
>> >> -----------------------------------------------------------------------------
>> >>
>> >> >From the above messages can anybody help me out in tracing the problem. Chris :-)
>> >>
>> >> Thanks in advance.
>> >>
>> >> Warm Regards
>> >>
>> >> Rajesh
>> >>
>> >> On Wed, Nov 01, 2000 at 11:23:14PM +0100, Christoph Plattner wrote:
>> >>
>> >> >
>> >> >This problem never should occur. The libraries `/sbin/init' needs, are
>> >> >libc.so.6 and ld-linux.so.2. Both libraries are in /lib and have to
>> >> >be on the root file system, which is mounted ! To be sure, copy the
>> >> >/etc/ld.so.cache file to your client path (/tftpboot/rajeshpc/root/etc,
>> >> >or whatever). Or you can use the ldconfig tool, which is able (in
>> >> >newer versions, check man page) to configure on an alternate root.
>> >> >If, for example, /tftpboot/rajeshpc/root is the path of the root file
>> >> >system, then run
>> >> >       ldconfig -r /tftpboot/rajeshpc/root
>> >> >so it sets up the paths
>> >> >       /tftpboot/rajeshpc/root/lib
>> >> >and
>> >> >       /tftpboot/rajeshpc/root/etc/ld.so.cache
>> >> >
>> >> >I hope this helps.
>> >> >
>> >> >Another important point !
>> >> >
>> >> >Check if the network configuration for the client is correct. On my
>> >> >first netboot attempts (one year ago), I had a small copied root
>> >> >file system from the server for the client. The problem was, that
>> >> >the setup for the ethernet was wrong (setup from server not from
>> >> >client !!). So the client was configured well over BOOTP, but
>> >> >when the system comes up via init, it also (re)initialized the
>> >> >ethernet wit the "wrong" setting and suddenly the client looses
>> >> >the contact to the server. The first reaction was the missed
>> >> >libraries, then blocking for file access and finally the error
>> >> >NFS error 111, which means `no access to server'.
>> >> >
>> >> >So here are some items to check again !
>> >> >
>> >> >
>> >> >Rajesh Fowkar wrote:
>> >> >>
>> >> >> Hi Chris,
>> >> >>
>> >> >> Initially I had copied the kernel in /tftpboot and the Diskless client was booting the kernel and than giving -13 error.
>> >> >>
>> >> >> Now I thought of copying the kernel to the actual workstation directory /tftpboot/rajeshpc and changed the dhcpd.conf file to point to /tftpboot/rajeshpc/kernel.
>> >> >>
>> >> >> Now the root filesystem is getting mounted but the init part gives error since the init is I think dynamically linked. The diskless node gives a list of errors like
>> >> >>
>> >> >> killing process getty
>> >> >> killing process init etc.
>> >> >> error in mounting shared libraries etc.
>> >> >>
>> >> >> I am not getting the login prompt. All these error are reapeating on the screen.
>> >> >>
>> >> >> How can I solve this problem to get the login prompt on the Diskless client.
>> >> >>
>> >> >> Thanks in advance.
>> >> >>
>> >> >> Warm Regards
>> >> >>
>> >> >> Rajesh
>> >> >>
>> >> >> >
>> >> >> >First of all test the mount point on your local linux server:
>> >> >> >
>> >> >> >       bash# mount -t nfs your_server:/tftpboot/xxxxxx /mnt/tmp
>> >> >> >
>> >> >> >(Of course, you must have a /mnt/tmp directory !)
>> >> >> >
>> >> >> >Further try following to nfs root for the diskless client:
>> >> >> >2 ways:
>> >> >> >
>> >> >> >ONE:   Use bootloader GNU Grub. Compile it with network support
>> >> >> >       (uses etherboot network drivers !!)
>> >> >> >       Install boot loader to disk and boot from disk (on client)
>> >> >> >       Now you are at the GRUB prompt:
>> >> >> >
>> >> >> >       grub> bootp
>> >> >> >
>> >> >> >       (Now you see the server and client address)
>> >> >> >
>> >> >> >       grub> root (nd)
>> >> >> >
>> >> >> >       (Use the network connction as root for the boot loader (not linux !)
>> >> >> >
>> >> >> >       grub> kernel /tftpboot/xxxxxx/vmlinuz root=/dev/nfs
>> >> >> >                       nfsroot=ip_of_server:/tftpboot/xxxx/root ip=........
>> >> >> >
>> >> >> >       (All in one line, then the command ...)
>> >> >> >
>> >> >> >       grub> boot
>> >> >> >
>> >> >> >The advantage here is to try the kernel without any image and to use the
>> >> >> >kernel parameter line for the boot (no bootp or any automatic for the
>> >> >> >first test (see /usr/src/linux/Documentation/nfsroot.txt,
>> >> >> >       /usr/doc/HOWTO/mini/NFS-Root, /usr/doc/HOWTO/BootPrompt-HOWTO).
>> >> >> >
>> >> >> >The second way is to boot via ldlinux.sys (syslinux.cfg), like on the
>> >> >> >RedHat boot disk (!). Here in syslinux.cfg the kernel parameter line can
>> >> >> >also be added.
>> >> >> >
>> >> >> >I prefer GRUB for all those experiements. I added support for diskless
>> >> >> >boot of GRUB itself. So the boot image is nbgrub, and you never had to
>> >> >> >use mknbi(-linux) any more !!!
>> >> >> >With the use of the grub config file (menu.lst) you can also
>> >> >> >automatcally
>> >> >> >boot the linux, everything diskless !! (I use this at home !)
>> >> >> >
>> >> >> >With friendly regards
>> >> >> >
>> >> >> >       Christoph plattner
>> >> >> >
>> 
>> ===========================================================================
>> This Mail was sent to netboot mailing list by:
>> Rajesh Fowkar <Mail@dempos.com>
>> To get help about this list, send a mail with 'help' as the only string in
>> it's body to majordomo@baghira.han.de. If you have problems with this list,
>> send a mail to netboot-owner@baghira.han.de.
>
>-- 
>  +--------V--------+	Christoph.Plattner@alcatel.at
>  |  A L C A T E L  |	-----------------------------
>  +-----------------+	Phone: +43 1 27722 3706	
>         T A S		Fax:   +43 1 27722 3955
>===========================================================================
>This Mail was sent to netboot mailing list by:
>Christoph Plattner <christoph.plattner@alcatel.at>
>To get help about this list, send a mail with 'help' as the only string in
>it's body to majordomo@baghira.han.de. If you have problems with this list,
>send a mail to netboot-owner@baghira.han.de.

===========================================================================
This Mail was sent to netboot mailing list by:
Rajesh Fowkar <Mail@dempos.com>
To get help about this list, send a mail with 'help' as the only string in
it's body to majordomo@baghira.han.de. If you have problems with this list,
send a mail to netboot-owner@baghira.han.de.



For requests or suggestions regarding this mailing list archive please write to netboot@gkminix.han.de.