Hi all,
Is it possible to take the 686 CentOS 5, install that in virtual machine, install a custom kernel that is the 486 flavor, take that image and put it on a 8G CF card - insert that into a small form factor 486 class machine and have that work?
I am hoping the extra instructions in a 686 (MMX, SEE etc) are not used by any libraries that I would be using (just standard stuff) and the new 486 compiled kernel would not be using those instructions either.
Can this possible work? I am hoping to not use debian i386 basically.
Thanks,
Jerry
When you create a virtual machine don't you have a choice of what type of processor you want and so forth?
That would be the clean way to do it would it not?
On Tue, 8 Dec 2009, Alan McKay wrote:
When you create a virtual machine don't you have a choice of what type of processor you want and so forth?
nope
That would be the clean way to do it would it not?
It may be one way to do it, but that is not how xen or kvm are ordinarily set up under CentOS -- qemu has (had) the hooks to simulate the missing opcodes of some arch's, but at a performance penalty
-- Russ herrold
Jerry Geis wrote:
Hi all,
Is it possible to take the 686 CentOS 5, install that in virtual machine, install a custom kernel that is the 486 flavor, take that image and put it on a 8G CF card - insert that into a small form factor 486 class machine and have that work?
I am hoping the extra instructions in a 686 (MMX, SEE etc) are not used by any libraries that I would be using (just standard stuff) and the new 486 compiled kernel would not be using those instructions either.
Can this possible work? I am hoping to not use debian i386 basically.
first, you'd need to reconfigure that linux running in a VM to run in whatever hardware your embedded platform provides (network, disk, etc).
second, you'd not only need a new kernel, but also any usermode packages that are -i686 or -i586 would need recompiling. there's more than you might think.
check out http://www.damnsmalllinux.org/ that will run on nearly -anything- with very minimal resources. the standard version fits on a 50MB live mini-CD (those credit card shaped things).
It may be one way to do it, but that is not how xen or kvm are ordinarily set up under CentOS -- qemu has (had) the hooks to simulate the missing opcodes of some arch's, but at a performance penalty
if there is a way to "simulate" missing opcodes in the kernel - that would be great also. I dont care if there is a performace hit.
I am looking for a way to run the 686 centos on a 486 machine. I was hoping I could just recompile the kernel as 486 and any libraries would not be using MMX/SSE etc...
Looking for any solution.
Jerry
CentOS Community,
I need help with a CentOS DHCP server.
I have a simple 32bit CentOS 5.3 server running on an Intel chip server in a lab environment with two NIC's.
Interfaces: eth0 - Is connected to the Internet using a static public IP address. eth1 - Is connected to a private 10.1.1.0/24 LAN with no other access to the web. Runs DHCP to the internal client systems. Is the default gateway for all LAN traffic to the Internet. Runs iptables as the firewall between the LAN and the Internet.
On eth1 DHCP was running with no problems for some time. This lab system sat for months untouched and then we revisited this product/project only to find that DHCP would not start. It gave us this following error:
Failed to start dhcpd : Internet Systems Consortium DHCP Server V3.1.3 Copyright 2004-2009 Internet Systems Consortium. All rights reserved. For info, please visit https://www.isc.org/software/dhcp/ Wrote 0 leases to leases file. Listening on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Sending on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Can't bind to dhcp address: Address already in use Please make sure there is no other dhcp server running and that there's no entry for dhcp or bootp in /etc/inetd.conf. Also make sure you are not running HP JetAdmin software, which includes a bootp server.
There is no other DHCP server on this LAN or on the public /30 that eth0 connects to (not that eth0 would impact my internal LAN).
I saw there were ofcourse many systems updates for CentOS and thought that a might resolve. It did not.
I then downloaded many versions of ISC's DHCP and compile and tried each of them from source code. This problems still exists. I have tried even the very simplest of dhcp.conf files and DHCP will still not start. Have I found a bug in the ISC DHCP code? Unlikely. I hope that one of you has run into this before and can help me out. Thanks greatly in advance.
Respectfully,
Larry Kemp Network Engineer U.S. Metropolitan Telecom, LLC Bonita Springs FL USA
On Tue, Dec 8, 2009 at 3:03 PM, Kemp, Larry Larry.Kemp@usmetrotel.com wrote:
CentOS Community,
I need help with a CentOS DHCP server.
I have a simple 32bit CentOS 5.3 server running on an Intel chip server in a lab environment with two NIC's.
Interfaces: eth0 - Is connected to the Internet using a static public IP address. eth1 - Is connected to a private 10.1.1.0/24 LAN with no other access to the web. Runs DHCP to the internal client systems. Is the default gateway for all LAN traffic to the Internet. Runs iptables as the firewall between the LAN and the Internet.
On eth1 DHCP was running with no problems for some time. This lab system sat for months untouched and then we revisited this product/project only to find that DHCP would not start. It gave us this following error:
Failed to start dhcpd : Internet Systems Consortium DHCP Server V3.1.3 Copyright 2004-2009 Internet Systems Consortium. All rights reserved. For info, please visit https://www.isc.org/software/dhcp/ Wrote 0 leases to leases file. Listening on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Sending on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Can't bind to dhcp address: Address already in use Please make sure there is no other dhcp server running and that there's no entry for dhcp or bootp in /etc/inetd.conf. Also make sure you are not running HP JetAdmin software, which includes a bootp server.
<snip>
The error message makes me think another dhcpd daemon is running. Have you checked for this with 'ps aux | grep dhcpd' or similar?
Since you've done updates, I'd assume you've rebooted the server, so it seems strange this might be the issue. Could possibly check to see if there is indeed another dhcpd daemon running. Something along the lines of 'netstat -patun | grep dhcp' or look for anything listening on UDP:68, iirc.
Someone else will probably be able to provide better help.
-jonathan
On Tue, 2009-12-08 at 16:03 -0500, Kemp, Larry wrote:
CentOS Community,
I need help with a CentOS DHCP server.
I have a simple 32bit CentOS 5.3 server running on an Intel chip server in a lab environment with two NIC's.
Interfaces: eth0 - Is connected to the Internet using a static public IP address. eth1 - Is connected to a private 10.1.1.0/24 LAN with no other access to the web. Runs DHCP to the internal client systems. Is the default gateway for all LAN traffic to the Internet. Runs iptables as the firewall between the LAN and the Internet.
On eth1 DHCP was running with no problems for some time. This lab system sat for months untouched and then we revisited this product/project only to find that DHCP would not start. It gave us this following error:
Failed to start dhcpd : Internet Systems Consortium DHCP Server V3.1.3 Copyright 2004-2009 Internet Systems Consortium. All rights reserved. For info, please visit https://www.isc.org/software/dhcp/ Wrote 0 leases to leases file. Listening on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Sending on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Can't bind to dhcp address: Address already in use Please make sure there is no other dhcp server running and that there's no entry for dhcp or bootp in /etc/inetd.conf. Also make sure you are not running HP JetAdmin software, which includes a bootp server.
There is no other DHCP server on this LAN or on the public /30 that eth0 connects to (not that eth0 would impact my internal LAN).
I'm just guessing here, but I think that this message is telling you that something else is bound to that interface on port 67 ( DHCP server port ) which occasionally can happen by chance.
Try lsof like this ( as root, of course ):
lsof -i -Pn | grep :67
This should show you what has grabbed port 67 and it may be something you can stop and restart to get a different ( random ) port assignment.
Like I said, this is just a guess.
I saw there were ofcourse many systems updates for CentOS and thought that a might resolve. It did not.
I then downloaded many versions of ISC's DHCP and compile and tried each of them from source code. This problems still exists. I have tried even the very simplest of dhcp.conf files and DHCP will still not start. Have I found a bug in the ISC DHCP code? Unlikely. I hope that one of you has run into this before and can help me out. Thanks greatly in advance.
Respectfully,
Larry Kemp Network Engineer U.S. Metropolitan Telecom, LLC Bonita Springs FL USA _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Yep. Dnsmasq was parked on 67. Gonna have to "yum remove" him. Big thanks guys.
LK
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Ron Loftin Sent: Tuesday, December 08, 2009 4:13 PM To: CentOS mailing list Subject: Re: [CentOS] CentOS DHCP Server
On Tue, 2009-12-08 at 16:03 -0500, Kemp, Larry wrote:
CentOS Community,
I need help with a CentOS DHCP server.
I have a simple 32bit CentOS 5.3 server running on an Intel chip server in a lab environment with two NIC's.
Interfaces: eth0 - Is connected to the Internet using a static public IP address. eth1 - Is connected to a private 10.1.1.0/24 LAN with no other access to the web. Runs DHCP to the internal client systems. Is the default gateway for all LAN traffic to the Internet. Runs iptables as the firewall between the LAN and the Internet.
On eth1 DHCP was running with no problems for some time. This lab system sat for months untouched and then we revisited this product/project only to find that DHCP would not start. It gave us this following error:
Failed to start dhcpd : Internet Systems Consortium DHCP Server V3.1.3 Copyright 2004-2009 Internet Systems Consortium. All rights reserved. For info, please visit https://www.isc.org/software/dhcp/ Wrote 0 leases to leases file. Listening on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Sending on LPF/eth1/00:50:ba:c0:43:c7/10.1.1/24 Can't bind to dhcp address: Address already in use Please make sure there is no other dhcp server running and that there's no entry for dhcp or bootp in /etc/inetd.conf. Also make sure you are not running HP JetAdmin software, which includes a bootp server.
There is no other DHCP server on this LAN or on the public /30 that eth0 connects to (not that eth0 would impact my internal LAN).
I'm just guessing here, but I think that this message is telling you that something else is bound to that interface on port 67 ( DHCP server port ) which occasionally can happen by chance.
Try lsof like this ( as root, of course ):
lsof -i -Pn | grep :67
This should show you what has grabbed port 67 and it may be something you can stop and restart to get a different ( random ) port assignment.
Like I said, this is just a guess.
I saw there were ofcourse many systems updates for CentOS and thought that a might resolve. It did not.
I then downloaded many versions of ISC's DHCP and compile and tried each of them from source code. This problems still exists. I have tried even the very simplest of dhcp.conf files and DHCP will still not start. Have I found a bug in the ISC DHCP code? Unlikely. I hope that one of you has run into this before and can help me out. Thanks greatly in advance.
Respectfully,
Larry Kemp Network Engineer U.S. Metropolitan Telecom, LLC Bonita Springs FL USA _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Tue, Dec 8, 2009 at 4:35 PM, Kemp, Larry Larry.Kemp@usmetrotel.com wrote:
Yep. Dnsmasq was parked on 67. Gonna have to "yum remove" him. Big thanks guys.
better be sure first that it is not also acting as a DNS server since it does both
Dnsmasq...good lightweight program that works in many instances. This was a very good lesson for me to first do a less on /etc/services before I bother the group with questions (noob mistake). I just configured BIND to handle my DNS requests. This particular box is a hybrid/appliance of sorts that we are working on in a lab. It functions as a firewall between the Internet and a private /24 LAN running iptables. It runs Aterisk for internal/external SIP communications with a hacked up interface of Spencer's code to have a branded look. Internally it runs DHCP for phones and PC's and DNS for internal LAN and Internet resolution of hosts. And Apache an TFTP for booting of SIP-phone-filesystems and logos on the phones. Probably a much better/faster/lighter way to do it all but it serves test purposes for now. Thanks all for the direction.
LK
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Alan McKay Sent: Tuesday, December 08, 2009 4:45 PM To: CentOS mailing list Subject: Re: [CentOS] CentOS DHCP Server
On Tue, Dec 8, 2009 at 4:35 PM, Kemp, Larry Larry.Kemp@usmetrotel.com wrote:
Yep. Dnsmasq was parked on 67. Gonna have to "yum remove" him. Big thanks guys.
better be sure first that it is not also acting as a DNS server since it does both
See if someone else is hanging onto port 67
[root@morrison ~]# lsof -i -P | grep :67 dnsmasq 23404 nobody 5u IPv4 46029087 UDP *:67
dnsmasq is a popular light-weight DHCP server
On Tuesday 08 December 2009 19:54:03 Jerry Geis wrote:
It may be one way to do it, but that is not how xen or kvm are ordinarily set up under CentOS -- qemu has (had) the hooks to simulate the missing opcodes of some arch's, but at a performance penalty
if there is a way to "simulate" missing opcodes in the kernel - that would be great also. I dont care if there is a performace hit.
I am looking for a way to run the 686 centos on a 486 machine. I was hoping I could just recompile the kernel as 486 and any libraries would not be using MMX/SSE etc...
I am no expert on this, but have a feeling that you would basically need to recompile every package that does not have an .i386 rpm.
And if you are about to recompile things, why not use gentoo or something like that?
Best, :-) Marko
At Tue, 08 Dec 2009 14:02:28 -0500 CentOS mailing list centos@centos.org wrote:
Hi all,
Is it possible to take the 686 CentOS 5, install that in virtual machine, install a custom kernel that is the 486 flavor, take that image and put it on a 8G CF card - insert that into a small form factor 486 class machine and have that work?
Will this small form factor 486 class machine be on the public internet network? If not (and thus not needing any current security patches, etc.), you might consider install RH 7.3 on it. RH 7.3 includes a *stock* i386 2.4 kernel (RedHat never built i486 kernels and I am not sure if anyone else bothered to either). I would also guess that the small form factor 486 class machine probably is NOT going to have modern I/O items, so you *probably* won't need any new kernel modules (eg SATA or fancy video or sound or network cards, etc.). I fould a set of RH 7.3 CDs on-line (*one* of the old mirrors still has it on line:
ftp://mirror.atlantic.net/pub/redhat-archive/7.3/
And I snarfed the ISOs to my server:
ftp://ftp.deepsoft.com/pub/RH7.3/iso/i386/
(I have an interst in using RH 7.3 on old '486 boxes and use them as model RR control nodes.)
I am hoping the extra instructions in a 686 (MMX, SEE etc) are not used by any libraries
Only some libraries (glibc for one, but it is available in several 'flavors', I think including a vanila i386 flavor).
that I would be using (just standard stuff) and the new 486 compiled kernel would not be using those instructions either.
Can this possible work? I am hoping to not use debian i386 basically.
Since various packages that would be installed on a 686 machine would be built for the 686 arch, you can't just copy the image of the 686 install onto a 486 system. You'll need to do a re-install to a 486 target. You can rebuild the kernel on the 686 (virtual or otherwise) and I believe all of the *.i686.rpm's are available as *.i386.rpm's. Somehow you will need to hack things at the install level to force a plain i386 install. You might need to hack anaconda and insert the i386 kernel or some such. Or do a careful downgrade of all of the i686 packages to i386 packages (in addition to the kernel itself).
Thanks,
Jerry _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Alan McKay -
Jerry Geis -
John R Pierce -
Jon Moore -
Kemp, Larry -
Marko Vojinovic -
R P Herrold -
Robert Heller -
Ron Loftin