2002-09-26 10:21:30

by Marco Schwarz

[permalink] [raw]
Subject: Serious Problems with diskless clients

Hi all,

my diskless clients have some severe problems on one of my servers.
Sometimes (right now most of the time) everything just hangs at the same place when
starting up the kernel. Here are the last messages I get (right before this
IP-Config is running and looks OK):

NET4: Unix domain sockets 1.0/SMP for Linux NET4.0
ds: no socket drivers loaded !
Looking up port of RPC 100003/2 on 192.168.0.235
portmap: server 192.168.0.235 mot responding, timed out !
Root-NFS: Unable to get nfsd port number from server, using default
Looking up port of RPC 100005/1 on 192.168.0.235
portmap: server 192.168.0.235 mot responding, timed out !
Root-NFS: Unable to get mountd port number from server, using default
mount: server 192.168.0.235 not responding, timed out
Root-NFS: Server returned error -5 while mounting /netclients/192.168.0.87
VFS: Unable to mount root fs via NFS, trying floppy
VFS: Insert root floppy and press ENTER

I am thinking right now that we have some problems with network hardware,
but maybe its a Software problem. Could someone tell me what the 'Looking up
port of RPC 100003/2 on 192.168.0.235' in kernel startup is doing an why it
could fail ?

We have Kernel 2.4.10 on both server and clients (I also tried 2.4.19, but
it changed nothing).

Thanks,
Marco Schwarz


2002-09-26 11:09:32

by Wakko Warner

[permalink] [raw]
Subject: Re: Serious Problems with diskless clients

> my diskless clients have some severe problems on one of my servers.
> Sometimes (right now most of the time) everything just hangs at the same place when
> starting up the kernel. Here are the last messages I get (right before this
> IP-Config is running and looks OK):
>
> NET4: Unix domain sockets 1.0/SMP for Linux NET4.0
> ds: no socket drivers loaded !
> Looking up port of RPC 100003/2 on 192.168.0.235
> portmap: server 192.168.0.235 mot responding, timed out !
> Root-NFS: Unable to get nfsd port number from server, using default
> Looking up port of RPC 100005/1 on 192.168.0.235
> portmap: server 192.168.0.235 mot responding, timed out !
> Root-NFS: Unable to get mountd port number from server, using default
> mount: server 192.168.0.235 not responding, timed out
> Root-NFS: Server returned error -5 while mounting /netclients/192.168.0.87
> VFS: Unable to mount root fs via NFS, trying floppy
> VFS: Insert root floppy and press ENTER
>
> I am thinking right now that we have some problems with network hardware,
> but maybe its a Software problem. Could someone tell me what the 'Looking up
> port of RPC 100003/2 on 192.168.0.235' in kernel startup is doing an why it
> could fail ?
>
> We have Kernel 2.4.10 on both server and clients (I also tried 2.4.19, but
> it changed nothing).

I have 2 diskless machines both tftping the kernel from the network and they
work just fine. Both are using kernel 2.4.19 vanalla with a small patch to
force ip=auto if nfs=/dev/nfs

[[email protected]:/] uname -a
Linux gohan 2.4.19 #1 SMP Tue Sep 3 13:02:36 EDT 2002 i686 unknown
[[email protected]:/] mount
rod:/tftpboot/gohan on / type nfs
(rw,intr,hard,rsize=8192,wsize=8192,intr,hard,rsize=8192,wsize=8192)
/proc on /proc type proc (rw)
devpts on /dev/pts type devpts (rw,gid=5,mode=620)
[[email protected]:/]

It's a very small patch to make that work w/o saying ip=auto on the command
line (I did this originally because I used a kernel disk w/o a boot loader
on some boxes at work)

--- net/ipv4/ipconfig-orig.c 2001-11-19 20:48:35.000000000 -0500
+++ net/ipv4/ipconfig.c 2001-11-19 20:56:21.000000000 -0500
@@ -1105,7 +1105,11 @@
proc_net_create("pnp", 0, pnp_get_info);
#endif /* CONFIG_PROC_FS */

- if (!ic_enable)
+ if (!ic_enable
+#if defined(IPCONFIG_DYNAMIC) && defined(CONFIG_ROOT_NFS)
+ && ROOT_DEV != MKDEV(UNNAMED_MAJOR, 255)
+#endif
+ )
return 0;

I did a cut'n'paste so I don't know if it will apply correctly but you get
the idea.

One thing I had problems with was using USB to mount the rootfs (usb hdd
actually) and I suspect the same with a USB nic (loading the kernel from
floppy).

--
Lab tests show that use of micro$oft causes cancer in lab animals

2002-09-26 12:58:28

by Denis Vlasenko

[permalink] [raw]
Subject: Re: Serious Problems with diskless clients

On 26 September 2002 08:26, Marco Schwarz wrote:
> Hi all,
>
> my diskless clients have some severe problems on one of my servers.
> Sometimes (right now most of the time) everything just hangs at the same
> place when starting up the kernel. Here are the last messages I get (right
> before this IP-Config is running and looks OK):
>
> NET4: Unix domain sockets 1.0/SMP for Linux NET4.0
> ds: no socket drivers loaded !
> Looking up port of RPC 100003/2 on 192.168.0.235
> portmap: server 192.168.0.235 mot responding, timed out !

Hook another box to the same network segment and run
ping or mtr to 192.168.0.235 and to the booting box.
Maybe your net drops packets or otherwise misbehaves.

BTW, 2.4.10 is way too old.
I don't see "mot responding, timed out !" in 2.4.19
source, rather "not responding, timed out".
--
vda

2002-09-26 13:11:00

by Marco Schwarz

[permalink] [raw]
Subject: Re: Serious Problems with diskless clients

> On 26 September 2002 08:26, Marco Schwarz wrote:
> > Hi all,
> >
> > my diskless clients have some severe problems on one of my servers.
> > Sometimes (right now most of the time) everything just hangs at the same
> > place when starting up the kernel. Here are the last messages I get
> (right
> > before this IP-Config is running and looks OK):
> >
> > NET4: Unix domain sockets 1.0/SMP for Linux NET4.0
> > ds: no socket drivers loaded !
> > Looking up port of RPC 100003/2 on 192.168.0.235
> > portmap: server 192.168.0.235 mot responding, timed out !
>
> Hook another box to the same network segment and run
> ping or mtr to 192.168.0.235 and to the booting box.
> Maybe your net drops packets or otherwise misbehaves.
>
> BTW, 2.4.10 is way too old.
> I don't see "mot responding, timed out !" in 2.4.19
> source, rather "not responding, timed out".
> --
> vda
>

"mot responding" is just a typo, I had to type all the messages from screen
;-)

I already tried pinging, works in both directions. On the server I start
portmap now with 'portmap -v' and I am able to see the requests from the client:


'connect from 192.168.0.87 to getport(nsf)'
'connect from 192.168.0.87 to getport(mountd)'

I also see some messages from Portmap which look like this:

'connect from 192.168.0.87 to dump()'

I also have to note that I have 2 NICs in this server, one with adress
153.95.240.x and one with 192.168.0.x.

Problems seem to occur only on the 192.168.0.x network (I already
interchanged adresses between cards, no effect).

BTW: Is there a newer version of portmap than 5.1 ? I wonder if this is
maybe related to portmap ...

Regards,
Marco

2002-09-26 21:34:15

by Daniel Egger

[permalink] [raw]
Subject: Re: Serious Problems with diskless clients

Am Don, 2002-09-26 um 13.18 schrieb Wakko Warner:

> --- net/ipv4/ipconfig-orig.c 2001-11-19 20:48:35.000000000 -0500
> +++ net/ipv4/ipconfig.c 2001-11-19 20:56:21.000000000 -0500
> @@ -1105,7 +1105,11 @@
> proc_net_create("pnp", 0, pnp_get_info);
> #endif /* CONFIG_PROC_FS */
>
> - if (!ic_enable)
> + if (!ic_enable
> +#if defined(IPCONFIG_DYNAMIC) && defined(CONFIG_ROOT_NFS)
> + && ROOT_DEV != MKDEV(UNNAMED_MAJOR, 255)
> +#endif
> + )
> return 0;

This together with the nfs-root-path-patch would be a nice addition to
the kernels I think, I find myself forgetting to add a few options every
now and then which is really nasty time-wise.

--
Servus,
Daniel


Attachments:
signature.asc (189.00 B)
Dies ist ein digital signierter Nachrichtenteil