petro at corp
Oct 29, 2003, 4:40 PM
Post #1 of 3
(1421 views)
Permalink
|
|
pxeboot.0 file fails to transfer?
|
|
On Wed, Oct 29, 2003 at 12:41:29AM -0500, Tom Diehl wrote:
> On Tue, 28 Oct 2003, Petro wrote:
> > On Tue, Oct 28, 2003 at 03:53:56PM -0800, H. Peter Anvin wrote:
> > > Petro wrote:
> > > > Is this the right place to ask questions about PXE and booting linux via
> > > > PXE?
> > > At least as long as you're using PXELINUX to do it...
> > Hmmm...
> > Well, it's the default c... that comes with Redhat 9, and syslinux
> > 2.00 is installed.
> > I'm running /usr/sbin/pxe and the (apparently unsupported) mtftpd
> > server (why does Redhat do this silly sh**?).
> They no longer do this. Even in redhat 9 you do not need the pxe package if you
> setup a dhcp server and tftp server.
> > I suspect my problem is before I even get to that point though. When
> > I'm trying to get a machine to boot I get this in my logs:
> > PXE[2723]: PxeService: WinSock error 0x58 on UDP recvfrom() on Port 67
> > Any ideas? Or do I need to rip out all this Redhat stuff and start
> > over with the right software?
> Ditch the pxe package and setup a dhcp server, tftp server, and use syslinux.
> Simple to setup and simple to use. See the syslinux home page for details.
> Once you have read the docs if you still have questions you can ask here.
Ok, did that.
Using tftp-ha:
[root [at] englab root]# /usr/sbin/in.tftpd -V
tftp-hpa 0.32, with remap, with tcpwrappers
I can successfull pull pxelinux.0 to my desktop (off the eth3
interface), but I cannot seem to get it off the server:
Here is the syslog going to my desk:
Oct 30 00:05:37 englab1 in.tftpd[3370]: RRQ from 10.128.1.234 filename /pxelinux.0
Oct 30 00:05:38 englab1 in.tftpd[3370]: tftpd: read(ack): Connection refused
pxelinux.0 successfully is copied over (md5sums match).
Here is the console from a pxeboot:
Intel LANDesk (R) Service Agent II, version 0.99c
Copyright (C) 1997 Intel Corporation. All rights reserved.
DHCP MAC ADDR: 00 D0 B7 88 01 52
PXE-EA1: No PXE server found, using standard boot file.
IP ADDR: 10.150.0.51
PXE-E32: TFTP open timeout.
PXE-E32: TFTP open timeout.
PXE-M0F: Exiting LANDesk (R) Service Agent II
And the syslogs:
Oct 29 16:28:54 englab1 dhcpd: DHCPDISCOVER from 00:d0:b7:88:01:52 via eth0
Oct 29 16:28:54 englab1 dhcpd: DHCPOFFER on 10.150.0.51 to 00:d0:b7:88:01:52 via eth0
Oct 29 16:28:54 englab1 dhcpd: DHCPREQUEST for 10.150.0.51 (10.150.0.1) from 00:d0:b7:88:01:52 via eth0
Oct 29 16:28:54 englab1 dhcpd: DHCPACK on 10.150.0.51 to 00:d0:b7:88:01:52 via eth0
Oct 30 00:28:58 englab1 in.tftpd[3379]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:29:02 englab1 in.tftpd[3380]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:29:10 englab1 in.tftpd[3381]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:29:22 englab1 in.tftpd[3382]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:29:38 englab1 in.tftpd[3383]: RRQ from 10.150.0.51 filename /pxelinux.0
And the logs from an attempt:
Oct 30 00:33:18 englab1 in.tftpd[3384]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:33:18 englab1 in.tftpd[3384]: tftpd: read(ack): Connection refused
Oct 30 00:33:23 englab1 in.tftpd[3385]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:33:23 englab1 in.tftpd[3385]: tftpd: read(ack): Connection refused
Oct 30 00:33:28 englab1 in.tftpd[3386]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:33:28 englab1 in.tftpd[3386]: tftpd: read(ack): Connection refused
Oct 30 00:33:33 englab1 in.tftpd[3387]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:33:33 englab1 in.tftpd[3387]: tftpd: read(ack): Connection refused
Oct 30 00:33:38 englab1 in.tftpd[3388]: RRQ from 10.150.0.51 filename /pxelinux.0
Oct 30 00:33:38 englab1 in.tftpd[3388]: tftpd: read(ack): Connection refused
In this latter case the pxelinux.0 file is not copied over.
tftpd is being started as in.tftpd from xinetd with these args:
server_args = -cvvp -a 10.150.0.1:69 -s /tftpboot
Which really should mean that my desktop can't get to the machine at
all, since it's coming in on the 10.128.x.x interface.
It's entirely possible I've got the server severely misconfigured
here--We're in the process of evaluating whether to switch from Debian
to Redhat, and I've not yet found were the folks at Redhat stick all of
their configuration stuff.
--
Petro [at] corp ccpetro at vtext.com [sms]
Unix Administrator 2766480 at skytel.com [pager]
Vendio Service Inc. (650) 793-1650 [cell]
|
