1:42 p.m.
Hi all,
I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
night.
Interestingly, when I run cobbler sync, only machines that have netboot
enabled get entries in my dhcpd.conf.
The offending line(s) appear to be in
/usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
164-165
----
if not interface["netboot_enabled"]:
continue
----
(This is in the section that generates the ISC dhcpd config file).
I commented those lines out to get my server back, but I was curious as
to why that was there in the first place.
Any insight would be appreciated!
Best,
Dan
3:18 p.m.
On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu> wrote:
Hi all,
I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
night.
Interestingly, when I run cobbler sync, only machines that have netboot
enabled get entries in my dhcpd.conf.
The offending line(s) appear to be in
/usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
164-165
----
if not interface["netboot_enabled"]:
continue
----
(This is in the section that generates the ISC dhcpd config file).
I commented those lines out to get my server back, but I was curious as
to why that was there in the first place.
Any insight would be appreciated!
It appears that this was added to prevent ppc64 reboot loops.
https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
I know that we had the potential for reboot loops even on regular x86
hardware, but we dealt with it without code change. Our systems boot
harddrive then netboot. So on initial install they netboot, once
there is an OS they regular boot. If you need to re-install you
either use koan or function key to select network boot at boot time.
That being said, our systems also turn static post-install, so this
change would not affect us. However, if we ran off DHCP period we
would have the same issue you are experiencing with this.
Maybe the better check would be:
if not interface["netboot_enabled"] and interface['static']:
continue
?
-greg
3:50 p.m.
On Fri, Jun 8, 2012 at 3:18 PM, Greg Swift <gregswift(a)gmail.com> wrote:
On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu>
wrote:
> Hi all,
>
> I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
> night.
>
> Interestingly, when I run cobbler sync, only machines that have netboot
> enabled get entries in my dhcpd.conf.
>
> The offending line(s) appear to be in
> /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
> 164-165
>
> ----
> if not interface["netboot_enabled"]:
> continue
> ----
>
> (This is in the section that generates the ISC dhcpd config file).
>
>
> I commented those lines out to get my server back, but I was curious as
> to why that was there in the first place.
>
> Any insight would be appreciated!
It appears that this was added to prevent ppc64 reboot loops.
https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
I know that we had the potential for reboot loops even on regular x86
hardware, but we dealt with it without code change. Our systems boot
harddrive then netboot. So on initial install they netboot, once
there is an OS they regular boot. If you need to re-install you
either use koan or function key to select network boot at boot time.
That being said, our systems also turn static post-install, so this
change would not affect us. However, if we ran off DHCP period we
would have the same issue you are experiencing with this.
Maybe the better check would be:
if not interface["netboot_enabled"] and interface['static']:
continue
?
also, might be good to file it as a bug on github?
4:12 p.m.
On 06/08/2012 03:50 PM, Greg Swift wrote:
On Fri, Jun 8, 2012 at 3:18 PM, Greg Swift
<gregswift(a)gmail.com> wrote:
> On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu> wrote:
>> Hi all,
>>
>> I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
>> night.
>>
>> Interestingly, when I run cobbler sync, only machines that have netboot
>> enabled get entries in my dhcpd.conf.
>>
>> The offending line(s) appear to be in
>> /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
>> 164-165
>>
>> ----
>> if not interface["netboot_enabled"]:
>> continue
>> ----
>>
>> (This is in the section that generates the ISC dhcpd config file).
>>
>>
>> I commented those lines out to get my server back, but I was curious as
>> to why that was there in the first place.
>>
>> Any insight would be appreciated!
>
> It appears that this was added to prevent ppc64 reboot loops.
> https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
>
> I know that we had the potential for reboot loops even on regular x86
> hardware, but we dealt with it without code change. Our systems boot
> harddrive then netboot. So on initial install they netboot, once
> there is an OS they regular boot. If you need to re-install you
> either use koan or function key to select network boot at boot time.
>
> That being said, our systems also turn static post-install, so this
> change would not affect us. However, if we ran off DHCP period we
> would have the same issue you are experiencing with this.
>
> Maybe the better check would be:
>
> if not interface["netboot_enabled"] and interface['static']:
> continue
>
> ?
also, might be good to file it as a bug on github?
Will do.
I've been meaning to get started contributing to the project anyways (we
use IPv6 heavily and have some enhancements to Cobbler in that
area)...this is a good push to get me off my rear!
Best,
Dan
4:35 p.m.
On 08.06.2012 [15:18:32 -0500], Greg Swift wrote:
On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu>
wrote:
> Hi all,
>
> I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
> night.
>
> Interestingly, when I run cobbler sync, only machines that have netboot
> enabled get entries in my dhcpd.conf.
>
> The offending line(s) appear to be in
> /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
> 164-165
>
> ----
> if not interface["netboot_enabled"]:
> continue
> ----
>
> (This is in the section that generates the ISC dhcpd config file).
>
>
> I commented those lines out to get my server back, but I was curious as
> to why that was there in the first place.
>
> Any insight would be appreciated!
It appears that this was added to prevent ppc64 reboot loops.
https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
I know that we had the potential for reboot loops even on regular x86
hardware, but we dealt with it without code change. Our systems boot
harddrive then netboot. So on initial install they netboot, once
there is an OS they regular boot. If you need to re-install you
either use koan or function key to select network boot at boot time.
That sounds ... not stand-alone :)
That being said, our systems also turn static post-install, so this
change would not affect us. However, if we ran off DHCP period we
would have the same issue you are experiencing with this.
Maybe the better check would be:
if not interface["netboot_enabled"] and interface['static']:
continue
?
Well, this actually perhaps points at my ignorance -- I think that if
you want Cobbler to manage DHCP, you mean to allow it to manage DHCP for
its purposes (which are transparent to yours). But if you want to have
DHCP be a persistent thing (unrelated to Cobbler's interactions), then
you probably should be managing it outside of Cobbler (at least, IMHO).
Or that's how I thought it was being used ... I hadn't considered the
case of using DHCP outside the netbooting case.
All of my machines are statically assigned IPs, and I only use DHCP on
installation. So, I think your above suggestion should work.
Thanks,
Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
4:40 p.m.
On 08.06.2012 [14:35:16 -0700], Nishanth Aravamudan wrote:
On 08.06.2012 [15:18:32 -0500], Greg Swift wrote:
> On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu> wrote:
> > Hi all,
> >
> > I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
> > night.
> >
> > Interestingly, when I run cobbler sync, only machines that have netboot
> > enabled get entries in my dhcpd.conf.
> >
> > The offending line(s) appear to be in
> > /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
> > 164-165
> >
> > ----
> > if not interface["netboot_enabled"]:
> > continue
> > ----
> >
> > (This is in the section that generates the ISC dhcpd config file).
> >
> >
> > I commented those lines out to get my server back, but I was curious as
> > to why that was there in the first place.
> >
> > Any insight would be appreciated!
>
> It appears that this was added to prevent ppc64 reboot loops.
> https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
>
> I know that we had the potential for reboot loops even on regular x86
> hardware, but we dealt with it without code change. Our systems boot
> harddrive then netboot. So on initial install they netboot, once
> there is an OS they regular boot. If you need to re-install you
> either use koan or function key to select network boot at boot time.
That sounds ... not stand-alone :)
> That being said, our systems also turn static post-install, so this
> change would not affect us. However, if we ran off DHCP period we
> would have the same issue you are experiencing with this.
>
> Maybe the better check would be:
>
> if not interface["netboot_enabled"] and interface['static']:
> continue
>
> ?
Well, this actually perhaps points at my ignorance -- I think that if
you want Cobbler to manage DHCP, you mean to allow it to manage DHCP for
its purposes (which are transparent to yours). But if you want to have
DHCP be a persistent thing (unrelated to Cobbler's interactions), then
you probably should be managing it outside of Cobbler (at least, IMHO).
Or that's how I thought it was being used ... I hadn't considered the
case of using DHCP outside the netbooting case.
And I will now affirm my own ignorance, after re-reading the main
Cobbler page:
"Cobbler can also optionally help with managing DHCP, DNS, and yum
package mirroring infrastructure -- in this regard, it is a more
generalized automation app, rather than just dealing specifically with
installations."
So, to that end, the above change is needed and I apologize for the
disruption.
Thanks,
Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
5:09 p.m.
On Fri, Jun 8, 2012 at 4:40 PM, Nishanth Aravamudan
<nacc(a)linux.vnet.ibm.com> wrote:
On 08.06.2012 [14:35:16 -0700], Nishanth Aravamudan wrote:
> On 08.06.2012 [15:18:32 -0500], Greg Swift wrote:
> > On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu> wrote:
> > > Hi all,
> > >
> > > I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
> > > night.
> > >
> > > Interestingly, when I run cobbler sync, only machines that have netboot
> > > enabled get entries in my dhcpd.conf.
> > >
> > > The offending line(s) appear to be in
> > > /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
> > > 164-165
> > >
> > > ----
> > > if not interface["netboot_enabled"]:
> > > continue
> > > ----
> > >
> > > (This is in the section that generates the ISC dhcpd config file).
> > >
> > >
> > > I commented those lines out to get my server back, but I was curious as
> > > to why that was there in the first place.
> > >
> > > Any insight would be appreciated!
> >
> > It appears that this was added to prevent ppc64 reboot loops.
> >
https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
> >
> > I know that we had the potential for reboot loops even on regular x86
> > hardware, but we dealt with it without code change. Our systems boot
> > harddrive then netboot. So on initial install they netboot, once
> > there is an OS they regular boot. If you need to re-install you
> > either use koan or function key to select network boot at boot time.
>
> That sounds ... not stand-alone :)
>
> > That being said, our systems also turn static post-install, so this
> > change would not affect us. However, if we ran off DHCP period we
> > would have the same issue you are experiencing with this.
> >
> > Maybe the better check would be:
> >
> > if not interface["netboot_enabled"] and interface['static']:
> > continue
> >
> > ?
>
> Well, this actually perhaps points at my ignorance -- I think that if
> you want Cobbler to manage DHCP, you mean to allow it to manage DHCP for
> its purposes (which are transparent to yours). But if you want to have
> DHCP be a persistent thing (unrelated to Cobbler's interactions), then
> you probably should be managing it outside of Cobbler (at least, IMHO).
> Or that's how I thought it was being used ... I hadn't considered the
> case of using DHCP outside the netbooting case.
And I will now affirm my own ignorance, after re-reading the main
Cobbler page:
"Cobbler can also optionally help with managing DHCP, DNS, and yum
package mirroring infrastructure -- in this regard, it is a more
generalized automation app, rather than just dealing specifically with
installations."
So, to that end, the above change is needed and I apologize for the
disruption.
it happens :) you meant well, and it is a nice change overall, as my
environment benefits from the concept, because now i don't care about
my boot order as much :)
-greg
10:10 p.m.
On Fri, Jun 8, 2012 at 5:09 PM, Greg Swift <gregswift(a)gmail.com> wrote:
On Fri, Jun 8, 2012 at 4:40 PM, Nishanth Aravamudan
<nacc(a)linux.vnet.ibm.com> wrote:
> On 08.06.2012 [14:35:16 -0700], Nishanth Aravamudan wrote:
>> On 08.06.2012 [15:18:32 -0500], Greg Swift wrote:
>> > On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu> wrote:
>> > > Hi all,
>> > >
>> > > I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
>> > > night.
>> > >
>> > > Interestingly, when I run cobbler sync, only machines that have
netboot
>> > > enabled get entries in my dhcpd.conf.
>> > >
>> > > The offending line(s) appear to be in
>> > > /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
>> > > 164-165
>> > >
>> > > ----
>> > > if not interface["netboot_enabled"]:
>> > > continue
>> > > ----
>> > >
>> > > (This is in the section that generates the ISC dhcpd config file).
>> > >
>> > >
>> > > I commented those lines out to get my server back, but I was curious
as
>> > > to why that was there in the first place.
>> > >
>> > > Any insight would be appreciated!
>> >
>> > It appears that this was added to prevent ppc64 reboot loops.
>> >
https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
>> >
>> > I know that we had the potential for reboot loops even on regular x86
>> > hardware, but we dealt with it without code change. Our systems boot
>> > harddrive then netboot. So on initial install they netboot, once
>> > there is an OS they regular boot. If you need to re-install you
>> > either use koan or function key to select network boot at boot time.
>>
>> That sounds ... not stand-alone :)
>>
>> > That being said, our systems also turn static post-install, so this
>> > change would not affect us. However, if we ran off DHCP period we
>> > would have the same issue you are experiencing with this.
>> >
>> > Maybe the better check would be:
>> >
>> > if not interface["netboot_enabled"] and
interface['static']:
>> > continue
>> >
>> > ?
>>
>> Well, this actually perhaps points at my ignorance -- I think that if
>> you want Cobbler to manage DHCP, you mean to allow it to manage DHCP for
>> its purposes (which are transparent to yours). But if you want to have
>> DHCP be a persistent thing (unrelated to Cobbler's interactions), then
>> you probably should be managing it outside of Cobbler (at least, IMHO).
>> Or that's how I thought it was being used ... I hadn't considered the
>> case of using DHCP outside the netbooting case.
>
> And I will now affirm my own ignorance, after re-reading the main
> Cobbler page:
>
> "Cobbler can also optionally help with managing DHCP, DNS, and yum
> package mirroring infrastructure -- in this regard, it is a more
> generalized automation app, rather than just dealing specifically with
> installations."
>
> So, to that end, the above change is needed and I apologize for the
> disruption.
it happens :) you meant well, and it is a nice change overall, as my
environment benefits from the concept, because now i don't care about
my boot order as much :)
-greg
Dan and Nishanth,
Did you try out the different conditional logic to see if it meets you needs?
1:27 a.m.
On 08.06.2012 [22:10:33 -0500], Greg Swift wrote:
On Fri, Jun 8, 2012 at 5:09 PM, Greg Swift
<gregswift(a)gmail.com> wrote:
> On Fri, Jun 8, 2012 at 4:40 PM, Nishanth Aravamudan
> <nacc(a)linux.vnet.ibm.com> wrote:
>> On 08.06.2012 [14:35:16 -0700], Nishanth Aravamudan wrote:
>>> On 08.06.2012 [15:18:32 -0500], Greg Swift wrote:
>>> > On Fri, Jun 8, 2012 at 1:42 PM, Dan Rao <dhananja(a)cs.wisc.edu>
wrote:
>>> > > Hi all,
>>> > >
>>> > > I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing)
last
>>> > > night.
>>> > >
>>> > > Interestingly, when I run cobbler sync, only machines that have
netboot
>>> > > enabled get entries in my dhcpd.conf.
>>> > >
>>> > > The offending line(s) appear to be in
>>> > > /usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py -
lines
>>> > > 164-165
>>> > >
>>> > > ----
>>> > > if not interface["netboot_enabled"]:
>>> > > continue
>>> > > ----
>>> > >
>>> > > (This is in the section that generates the ISC dhcpd config
file).
>>> > >
>>> > >
>>> > > I commented those lines out to get my server back, but I was
curious as
>>> > > to why that was there in the first place.
>>> > >
>>> > > Any insight would be appreciated!
>>> >
>>> > It appears that this was added to prevent ppc64 reboot loops.
>>> >
https://github.com/cobbler/cobbler/commit/bc76631ea158c76b1cdc8d3af28c9dc...
>>> >
>>> > I know that we had the potential for reboot loops even on regular x86
>>> > hardware, but we dealt with it without code change. Our systems boot
>>> > harddrive then netboot. So on initial install they netboot, once
>>> > there is an OS they regular boot. If you need to re-install you
>>> > either use koan or function key to select network boot at boot time.
>>>
>>> That sounds ... not stand-alone :)
>>>
>>> > That being said, our systems also turn static post-install, so this
>>> > change would not affect us. However, if we ran off DHCP period we
>>> > would have the same issue you are experiencing with this.
>>> >
>>> > Maybe the better check would be:
>>> >
>>> > if not interface["netboot_enabled"] and
interface['static']:
>>> > continue
>>> >
>>> > ?
>>>
>>> Well, this actually perhaps points at my ignorance -- I think that if
>>> you want Cobbler to manage DHCP, you mean to allow it to manage DHCP for
>>> its purposes (which are transparent to yours). But if you want to have
>>> DHCP be a persistent thing (unrelated to Cobbler's interactions), then
>>> you probably should be managing it outside of Cobbler (at least, IMHO).
>>> Or that's how I thought it was being used ... I hadn't considered
the
>>> case of using DHCP outside the netbooting case.
>>
>> And I will now affirm my own ignorance, after re-reading the main
>> Cobbler page:
>>
>> "Cobbler can also optionally help with managing DHCP, DNS, and yum
>> package mirroring infrastructure -- in this regard, it is a more
>> generalized automation app, rather than just dealing specifically with
>> installations."
>>
>> So, to that end, the above change is needed and I apologize for the
>> disruption.
>
> it happens :) you meant well, and it is a nice change overall, as my
> environment benefits from the concept, because now i don't care about
> my boot order as much :)
>
> -greg
Dan and Nishanth,
Did you try out the different conditional logic to see if it meets you
needs?
I'm testing it now in my environment, need to run a few more jobs, but
it seems fine here.
Thanks,
Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
9:17 a.m.
> Dan and Nishanth,
>
> Did you try out the different conditional logic to see if it meets you
> needs?
I'm testing it now in my environment, need to run a few more jobs, but
it seems fine here.
What I was actually thinking of doing to correct this is to not set
the next-server line if netboot-enabled=false. So basically we'd
revert Nishanth's change and then modify the dhcpd template:
#if $iface.netboot_enabled:
next-server $next_server;
#end if
This is already being set in modules/manage_isc.py:
interface["netboot_enabled"] = blended_system["netboot_enabled"]
So it should be a nice easy fix. Regardless, whatever we do I'll get
this pushed out as 2.2.3-2, as I didn't think through the impact this
might have to users that manage dhcpd with cobbler.
10:29 a.m.
On Sat, Jun 9, 2012 at 9:17 AM, James Cammarata <jimi(a)sngx.net> wrote:
>> Dan and Nishanth,
>>
>> Did you try out the different conditional logic to see if it meets you
>> needs?
>
> I'm testing it now in my environment, need to run a few more jobs, but
> it seems fine here.
What I was actually thinking of doing to correct this is to not set
the next-server line if netboot-enabled=false. So basically we'd
revert Nishanth's change and then modify the dhcpd template:
#if $iface.netboot_enabled:
next-server $next_server;
#end if
This is already being set in modules/manage_isc.py:
interface["netboot_enabled"] = blended_system["netboot_enabled"]
So it should be a nice easy fix. Regardless, whatever we do I'll get
this pushed out as 2.2.3-2, as I didn't think through the impact this
might have to users that manage dhcpd with cobbler.
Give this a try:
https://github.com/jimi1283/cobbler/tree/dhcp_fix
2:12 p.m.
On 06/09/2012 10:29 AM, James Cammarata wrote:
On Sat, Jun 9, 2012 at 9:17 AM, James Cammarata <jimi(a)sngx.net>
wrote:
>>> Dan and Nishanth,
>>>
>>> Did you try out the different conditional logic to see if it meets you
>>> needs?
>>
>> I'm testing it now in my environment, need to run a few more jobs, but
>> it seems fine here.
>
> What I was actually thinking of doing to correct this is to not set
> the next-server line if netboot-enabled=false. So basically we'd
> revert Nishanth's change and then modify the dhcpd template:
>
> #if $iface.netboot_enabled:
> next-server $next_server;
> #end if
>
> This is already being set in modules/manage_isc.py:
>
> interface["netboot_enabled"] = blended_system["netboot_enabled"]
>
> So it should be a nice easy fix. Regardless, whatever we do I'll get
> this pushed out as 2.2.3-2, as I didn't think through the impact this
> might have to users that manage dhcpd with cobbler.
Give this a try:
https://github.com/jimi1283/cobbler/tree/dhcp_fix
Works for me. Thanks!
Best,
Dan
10:11 p.m.
On 11.06.2012 [14:12:58 -0500], Dan Rao wrote:
On 06/09/2012 10:29 AM, James Cammarata wrote:
> On Sat, Jun 9, 2012 at 9:17 AM, James Cammarata <jimi(a)sngx.net> wrote:
>>>> Dan and Nishanth,
>>>>
>>>> Did you try out the different conditional logic to see if it meets you
>>>> needs?
>>>
>>> I'm testing it now in my environment, need to run a few more jobs, but
>>> it seems fine here.
>>
>> What I was actually thinking of doing to correct this is to not set
>> the next-server line if netboot-enabled=false. So basically we'd
>> revert Nishanth's change and then modify the dhcpd template:
>>
>> #if $iface.netboot_enabled:
>> next-server $next_server;
>> #end if
>>
>> This is already being set in modules/manage_isc.py:
>>
>> interface["netboot_enabled"] =
blended_system["netboot_enabled"]
>>
>> So it should be a nice easy fix. Regardless, whatever we do I'll get
>> this pushed out as 2.2.3-2, as I didn't think through the impact this
>> might have to users that manage dhcpd with cobbler.
>
> Give this a try:
>
> https://github.com/jimi1283/cobbler/tree/dhcp_fix
Works for me. Thanks!
I will have to test this out -- the issue for PowerPC is that a DHCP
response (which perhaps won't get sent if next-server is not set?)
caused the SMS menus to assume netbooting was the boot method and there
didn't seem to be any failover (fallback, perhaps) to disk if that
failed. But, like I said, I will test the patch and let you know,
hopefully tomorrow.
-Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
4:06 p.m.
On 11.06.2012 [20:11:38 -0700], Nishanth Aravamudan wrote:
On 11.06.2012 [14:12:58 -0500], Dan Rao wrote:
> On 06/09/2012 10:29 AM, James Cammarata wrote:
> > On Sat, Jun 9, 2012 at 9:17 AM, James Cammarata <jimi(a)sngx.net> wrote:
> >>>> Dan and Nishanth,
> >>>>
> >>>> Did you try out the different conditional logic to see if it meets
you
> >>>> needs?
> >>>
> >>> I'm testing it now in my environment, need to run a few more jobs,
but
> >>> it seems fine here.
> >>
> >> What I was actually thinking of doing to correct this is to not set
> >> the next-server line if netboot-enabled=false. So basically we'd
> >> revert Nishanth's change and then modify the dhcpd template:
> >>
> >> #if $iface.netboot_enabled:
> >> next-server $next_server;
> >> #end if
> >>
> >> This is already being set in modules/manage_isc.py:
> >>
> >> interface["netboot_enabled"] =
blended_system["netboot_enabled"]
> >>
> >> So it should be a nice easy fix. Regardless, whatever we do I'll get
> >> this pushed out as 2.2.3-2, as I didn't think through the impact this
> >> might have to users that manage dhcpd with cobbler.
> >
> > Give this a try:
> >
> > https://github.com/jimi1283/cobbler/tree/dhcp_fix
>
> Works for me. Thanks!
I will have to test this out -- the issue for PowerPC is that a DHCP
response (which perhaps won't get sent if next-server is not set?)
caused the SMS menus to assume netbooting was the boot method and there
didn't seem to be any failover (fallback, perhaps) to disk if that
failed. But, like I said, I will test the patch and let you know,
hopefully tomorrow.
So this change on its own is insufficient in my environment. Here is the
resulting bootlog:
TFTP BOOT ---------------------------------------------------
Server IP.....................9.3.191.134
Client IP.....................9.3.192.209
Gateway IP....................9.3.192.1
Subnet Mask...................255.255.255.0
( 1 ) Filename................./yaboot
TFTP Retries..................5
Block Size....................512
FINAL PACKET COUNT = 388
FINAL FILE SIZE = 198376 BYTES
Elapsed time since release of system processors: 18709 mins 13 secs
ipv6 = <0>
siaddr = <9.3.191.134>
file = <>
ciaddr = <9.3.192.209>
giaddr = <9.3.192.1>
bootp_retries = <5>
tftp_retries = <5>
addl_params = <512>
dhcpv6 = <<NULL>>
blksize = <<NULL>>
Try to netboot
TFTP BOOT ---------------------------------------------------
Server IP.....................9.3.191.134
Client IP.....................9.3.192.209
Gateway IP....................9.3.192.1
Subnet Mask...................255.255.255.0
( 1 ) Filename.................\etc\01-82-5c-60-4a-67-04
TFTP Retries..................5
Block Size....................512
1
2
3
!BA017021 !
Error, can't read config file
...
It tries a few other config file locations, but then eventually drops to
yaboot's prompt, with no kernels available, and if you hit enter there,
it just tries to load the conf file again.
FWIW, here was the generated stanza:
host generic7 {
hardware ethernet <AA:BB:CC:DD:EE:FF>;
fixed-address <ip>;
option host-name "<hostname>";
option subnet-mask 255.255.255.0;
option routers <gw>;
filename "/yaboot";
}
If I change your commit slightly, I do get what I need:
Do not generate either the next-server or filename sections of the DHCP
stanza if a server is not setup to netboot. This allows DHCP
environments to obtain IP addresses, but prevents netboot cycles on
PowerPC.
Signed-off-by: Nishanth Aravamudan <nacc(a)us.ibm.com>
diff --git a/templates/etc/dhcp.template b/templates/etc/dhcp.template
index d7a28ef..58926ab 100644
--- a/templates/etc/dhcp.template
+++ b/templates/etc/dhcp.template
@@ -50,6 +50,7 @@ group {
#if $iface.gateway:
option routers $iface.gateway;
#end if
+ #if $iface.netboot_enabled:
#if $iface.enable_gpxe:
if exists user-class and option user-class = "gPXE" {
filename
"http://$cobbler_server/cblr/svc/op/gpxe/system/$iface.owner";
@@ -61,7 +62,6 @@ group {
#end if
## Cobbler defaults to $next_server, but some users
## may like to use $iface.system.server for proxied setups
- #if $iface.netboot_enabled:
next-server $next_server;
#end if
## next-server $iface.next_server;
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
5:43 p.m.
So this change on its own is insufficient in my environment. Here is
the
resulting bootlog:
TFTP BOOT ---------------------------------------------------
Server IP.....................9.3.191.134
Client IP.....................9.3.192.209
Gateway IP....................9.3.192.1
Subnet Mask...................255.255.255.0
( 1 ) Filename................./yaboot
TFTP Retries..................5
Block Size....................512
FINAL PACKET COUNT = 388
FINAL FILE SIZE = 198376 BYTES
Elapsed time since release of system processors: 18709 mins 13 secs
ipv6 = <0>
siaddr = <9.3.191.134>
file = <>
ciaddr = <9.3.192.209>
giaddr = <9.3.192.1>
bootp_retries = <5>
tftp_retries = <5>
addl_params = <512>
dhcpv6 = <<NULL>>
blksize = <<NULL>>
Try to netboot
TFTP BOOT ---------------------------------------------------
Server IP.....................9.3.191.134
Client IP.....................9.3.192.209
Gateway IP....................9.3.192.1
Subnet Mask...................255.255.255.0
( 1 ) Filename.................\etc\01-82-5c-60-4a-67-04
TFTP Retries..................5
Block Size....................512
1
2
3
!BA017021 !
Error, can't read config file
...
It tries a few other config file locations, but then eventually drops to
yaboot's prompt, with no kernels available, and if you hit enter there,
it just tries to load the conf file again.
FWIW, here was the generated stanza:
host generic7 {
hardware ethernet <AA:BB:CC:DD:EE:FF>;
fixed-address <ip>;
option host-name "<hostname>";
option subnet-mask 255.255.255.0;
option routers <gw>;
filename "/yaboot";
}
If I change your commit slightly, I do get what I need:
Do not generate either the next-server or filename sections of the DHCP
stanza if a server is not setup to netboot. This allows DHCP
environments to obtain IP addresses, but prevents netboot cycles on
PowerPC.
Signed-off-by: Nishanth Aravamudan <nacc(a)us.ibm.com>
diff --git a/templates/etc/dhcp.template b/templates/etc/dhcp.template
index d7a28ef..58926ab 100644
--- a/templates/etc/dhcp.template
+++ b/templates/etc/dhcp.template
@@ -50,6 +50,7 @@ group {
#if $iface.gateway:
option routers $iface.gateway;
#end if
+ #if $iface.netboot_enabled:
#if $iface.enable_gpxe:
if exists user-class and option user-class = "gPXE" {
filename
"http://$cobbler_server/cblr/svc/op/gpxe/system/$iface.owner";
@@ -61,7 +62,6 @@ group {
#end if
## Cobbler defaults to $next_server, but some users
## may like to use $iface.system.server for proxied setups
- #if $iface.netboot_enabled:
next-server $next_server;
#end if
## next-server $iface.next_server;
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
It sounds like you need a new system (per-interface maybe?) setting to
prevent entries from being entered into the dhcpd.conf:
if not netboot_enabled and disable_dhcp_entry:
continue
disable_dhcp_entry would default to false.
5:50 p.m.
On Tue, Jun 12, 2012 at 5:43 PM, James Cammarata <jimi(a)sngx.net> wrote:
> So this change on its own is insufficient in my environment. Here
is the
> resulting bootlog:
>
> TFTP BOOT ---------------------------------------------------
> Server IP.....................9.3.191.134
> Client IP.....................9.3.192.209
> Gateway IP....................9.3.192.1
> Subnet Mask...................255.255.255.0
> ( 1 ) Filename................./yaboot
> TFTP Retries..................5
> Block Size....................512
> FINAL PACKET COUNT = 388
> FINAL FILE SIZE = 198376 BYTES
>
> Elapsed time since release of system processors: 18709 mins 13 secs
>
> ipv6 = <0>
> siaddr = <9.3.191.134>
> file = <>
> ciaddr = <9.3.192.209>
> giaddr = <9.3.192.1>
> bootp_retries = <5>
> tftp_retries = <5>
> addl_params = <512>
> dhcpv6 = <<NULL>>
> blksize = <<NULL>>
> Try to netboot
>
>
>
> TFTP BOOT ---------------------------------------------------
> Server IP.....................9.3.191.134
> Client IP.....................9.3.192.209
> Gateway IP....................9.3.192.1
> Subnet Mask...................255.255.255.0
> ( 1 ) Filename.................\etc\01-82-5c-60-4a-67-04
> TFTP Retries..................5
> Block Size....................512
> 1
> 2
> 3
> !BA017021 !
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Error, can't read config file
>
> ...
>
> It tries a few other config file locations, but then eventually drops to
> yaboot's prompt, with no kernels available, and if you hit enter there,
> it just tries to load the conf file again.
>
> FWIW, here was the generated stanza:
>
> host generic7 {
> hardware ethernet <AA:BB:CC:DD:EE:FF>;
> fixed-address <ip>;
> option host-name "<hostname>";
> option subnet-mask 255.255.255.0;
> option routers <gw>;
> filename "/yaboot";
> }
>
> If I change your commit slightly, I do get what I need:
>
>
>
> Do not generate either the next-server or filename sections of the DHCP
> stanza if a server is not setup to netboot. This allows DHCP
> environments to obtain IP addresses, but prevents netboot cycles on
> PowerPC.
>
> Signed-off-by: Nishanth Aravamudan <nacc(a)us.ibm.com>
>
> diff --git a/templates/etc/dhcp.template b/templates/etc/dhcp.template
> index d7a28ef..58926ab 100644
> --- a/templates/etc/dhcp.template
> +++ b/templates/etc/dhcp.template
> @@ -50,6 +50,7 @@ group {
> #if $iface.gateway:
> option routers $iface.gateway;
> #end if
> + #if $iface.netboot_enabled:
> #if $iface.enable_gpxe:
> if exists user-class and option user-class = "gPXE" {
> filename
"http://$cobbler_server/cblr/svc/op/gpxe/system/$iface.owner";
> @@ -61,7 +62,6 @@ group {
> #end if
> ## Cobbler defaults to $next_server, but some users
> ## may like to use $iface.system.server for proxied setups
> - #if $iface.netboot_enabled:
> next-server $next_server;
> #end if
> ## next-server $iface.next_server;
>
> --
> Nishanth Aravamudan <nacc(a)us.ibm.com>
> IBM Linux Technology Center
>
It sounds like you need a new system (per-interface maybe?) setting to
prevent entries from being entered into the dhcpd.conf:
if not netboot_enabled and disable_dhcp_entry:
continue
disable_dhcp_entry would default to false.
Isn't that just duplicating the function of checking the 'static' box?
-greg
6:11 p.m.
On 12.06.2012 [17:50:22 -0500], Greg Swift wrote:
On Tue, Jun 12, 2012 at 5:43 PM, James Cammarata
<jimi(a)sngx.net> wrote:
>> So this change on its own is insufficient in my environment. Here is the
>> resulting bootlog:
>>
>> TFTP BOOT ---------------------------------------------------
>> Server IP.....................9.3.191.134
>> Client IP.....................9.3.192.209
>> Gateway IP....................9.3.192.1
>> Subnet Mask...................255.255.255.0
>> ( 1 ) Filename................./yaboot
>> TFTP Retries..................5
>> Block Size....................512
>> FINAL PACKET COUNT = 388
>> FINAL FILE SIZE = 198376 BYTES
>>
>> Elapsed time since release of system processors: 18709 mins 13 secs
>>
>> ipv6 = <0>
>> siaddr = <9.3.191.134>
>> file = <>
>> ciaddr = <9.3.192.209>
>> giaddr = <9.3.192.1>
>> bootp_retries = <5>
>> tftp_retries = <5>
>> addl_params = <512>
>> dhcpv6 = <<NULL>>
>> blksize = <<NULL>>
>> Try to netboot
>>
>>
>>
>> TFTP BOOT ---------------------------------------------------
>> Server IP.....................9.3.191.134
>> Client IP.....................9.3.192.209
>> Gateway IP....................9.3.192.1
>> Subnet Mask...................255.255.255.0
>> ( 1 ) Filename.................\etc\01-82-5c-60-4a-67-04
>> TFTP Retries..................5
>> Block Size....................512
>> 1
>> 2
>> 3
>> !BA017021 !
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> Error, can't read config file
>>
>> ...
>>
>> It tries a few other config file locations, but then eventually drops to
>> yaboot's prompt, with no kernels available, and if you hit enter there,
>> it just tries to load the conf file again.
>>
>> FWIW, here was the generated stanza:
>>
>> host generic7 {
>> hardware ethernet <AA:BB:CC:DD:EE:FF>;
>> fixed-address <ip>;
>> option host-name "<hostname>";
>> option subnet-mask 255.255.255.0;
>> option routers <gw>;
>> filename "/yaboot";
>> }
>>
>> If I change your commit slightly, I do get what I need:
>>
>>
>>
>> Do not generate either the next-server or filename sections of the DHCP
>> stanza if a server is not setup to netboot. This allows DHCP
>> environments to obtain IP addresses, but prevents netboot cycles on
>> PowerPC.
>>
>> Signed-off-by: Nishanth Aravamudan <nacc(a)us.ibm.com>
>>
>> diff --git a/templates/etc/dhcp.template b/templates/etc/dhcp.template
>> index d7a28ef..58926ab 100644
>> --- a/templates/etc/dhcp.template
>> +++ b/templates/etc/dhcp.template
>> @@ -50,6 +50,7 @@ group {
>> #if $iface.gateway:
>> option routers $iface.gateway;
>> #end if
>> + #if $iface.netboot_enabled:
>> #if $iface.enable_gpxe:
>> if exists user-class and option user-class = "gPXE" {
>> filename
"http://$cobbler_server/cblr/svc/op/gpxe/system/$iface.owner";
>> @@ -61,7 +62,6 @@ group {
>> #end if
>> ## Cobbler defaults to $next_server, but some users
>> ## may like to use $iface.system.server for proxied setups
>> - #if $iface.netboot_enabled:
>> next-server $next_server;
>> #end if
>> ## next-server $iface.next_server;
>>
>> --
>> Nishanth Aravamudan <nacc(a)us.ibm.com>
>> IBM Linux Technology Center
>>
>
> It sounds like you need a new system (per-interface maybe?) setting to
> prevent entries from being entered into the dhcpd.conf:
>
> if not netboot_enabled and disable_dhcp_entry:
> continue
>
> disable_dhcp_entry would default to false.
Isn't that just duplicating the function of checking the 'static' box?
That's a fair point, and I had tested Greg's version, which just
appended a 'and interface["static"]' to the line I added, and it did
work for my environment, as all of my IPs are statically determined.
Thanks,
Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
12:51 p.m.
So since they both tested with success the patch and everything went
silent, I guess thats a yes?
On Wed, Jun 13, 2012 at 7:26 AM, James Cammarata <jimi(a)sngx.net> wrote:
> Isn't that just duplicating the function of checking the
'static' box?
I'll defer to you guys, if that original suggested patch works I'll
run with that.
_______________________________________________
cobbler mailing list
cobbler(a)lists.fedorahosted.org
https://fedorahosted.org/mailman/listinfo/cobbler
1:01 p.m.
On 15.06.2012 [12:51:16 -0500], Greg Swift wrote:
So since they both tested with success the patch and everything went
silent, I guess thats a yes?
On Wed, Jun 13, 2012 at 7:26 AM, James Cammarata <jimi(a)sngx.net> wrote:
>> Isn't that just duplicating the function of checking the 'static'
box?
>
> I'll defer to you guys, if that original suggested patch works I'll
> run with that.
Yeah, sorry, loaded with other work and forgot to reply. I have tested
the static patch and it worked fine in my environment.
Thanks,
Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
5:59 p.m.
On 12.06.2012 [17:43:28 -0500], James Cammarata wrote:
> So this change on its own is insufficient in my environment.
Here is the
> resulting bootlog:
>
> TFTP BOOT ---------------------------------------------------
> Server IP.....................9.3.191.134
> Client IP.....................9.3.192.209
> Gateway IP....................9.3.192.1
> Subnet Mask...................255.255.255.0
> ( 1 ) Filename................./yaboot
> TFTP Retries..................5
> Block Size....................512
> FINAL PACKET COUNT = 388
> FINAL FILE SIZE = 198376 BYTES
>
> Elapsed time since release of system processors: 18709 mins 13 secs
>
> ipv6 = <0>
> siaddr = <9.3.191.134>
> file = <>
> ciaddr = <9.3.192.209>
> giaddr = <9.3.192.1>
> bootp_retries = <5>
> tftp_retries = <5>
> addl_params = <512>
> dhcpv6 = <<NULL>>
> blksize = <<NULL>>
> Try to netboot
>
>
>
> TFTP BOOT ---------------------------------------------------
> Server IP.....................9.3.191.134
> Client IP.....................9.3.192.209
> Gateway IP....................9.3.192.1
> Subnet Mask...................255.255.255.0
> ( 1 ) Filename.................\etc\01-82-5c-60-4a-67-04
> TFTP Retries..................5
> Block Size....................512
> 1
> 2
> 3
> !BA017021 !
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Error, can't read config file
>
> ...
>
> It tries a few other config file locations, but then eventually drops to
> yaboot's prompt, with no kernels available, and if you hit enter there,
> it just tries to load the conf file again.
>
> FWIW, here was the generated stanza:
>
> host generic7 {
> hardware ethernet <AA:BB:CC:DD:EE:FF>;
> fixed-address <ip>;
> option host-name "<hostname>";
> option subnet-mask 255.255.255.0;
> option routers <gw>;
> filename "/yaboot";
> }
>
> If I change your commit slightly, I do get what I need:
>
>
>
> Do not generate either the next-server or filename sections of the DHCP
> stanza if a server is not setup to netboot. This allows DHCP
> environments to obtain IP addresses, but prevents netboot cycles on
> PowerPC.
>
> Signed-off-by: Nishanth Aravamudan <nacc(a)us.ibm.com>
>
> diff --git a/templates/etc/dhcp.template b/templates/etc/dhcp.template
> index d7a28ef..58926ab 100644
> --- a/templates/etc/dhcp.template
> +++ b/templates/etc/dhcp.template
> @@ -50,6 +50,7 @@ group {
> #if $iface.gateway:
> option routers $iface.gateway;
> #end if
> + #if $iface.netboot_enabled:
> #if $iface.enable_gpxe:
> if exists user-class and option user-class = "gPXE" {
> filename
"http://$cobbler_server/cblr/svc/op/gpxe/system/$iface.owner";
> @@ -61,7 +62,6 @@ group {
> #end if
> ## Cobbler defaults to $next_server, but some users
> ## may like to use $iface.system.server for proxied setups
> - #if $iface.netboot_enabled:
> next-server $next_server;
> #end if
> ## next-server $iface.next_server;
>
> --
> Nishanth Aravamudan <nacc(a)us.ibm.com>
> IBM Linux Technology Center
>
It sounds like you need a new system (per-interface maybe?) setting to
prevent entries from being entered into the dhcpd.conf:
if not netboot_enabled and disable_dhcp_entry:
continue
disable_dhcp_entry would default to false.
For my use-case, it would be a global thing (I have no use for DHCP
outside of net-booting), which would be nice to not have to twiddle
per-system, but is ok if I have to. I don't need it per-interface.
I'm not sure what that buys me as opposed to moving the conditional
above? If the intent was to have the lack of the next-server line
prevent netbooting, I'm not sure why it would matter if we have the
filename line or not correspondingly?
Thanks,
Nish
--
Nishanth Aravamudan <nacc(a)us.ibm.com>
IBM Linux Technology Center
10:36 a.m.
Some days ago i had questions about how to add only a host entry
in dhcp via cobbler and nothing more (so, for example, to serve a
printer via dhcp). The result was this: https://github.com/cobbler/cobbler/issues/192
and the fact, it is not possible to add a dhcp entry without
a profile and so on.
I read this discussion and i think it touch my wish ...
And for me it would be a nice feature. It is a good idea to have this on a wishlist?
(Or do you know a trick to do this already? Well, i not so familiar with the cobbler
source code ...)
Detlev
--
Detlev | Institut fuer Mikroelektronische Systeme
Habicht | D-30167 Hannover +49 511 76219662 habicht(a)ims.uni-hannover.de
--------+-------- Handy +49 172 5415752 ---------------------------
Am 09.06.2012 um 16:17 schrieb James Cammarata:
>> Dan and Nishanth,
>>
>> Did you try out the different conditional logic to see if it meets you
>> needs?
>
> I'm testing it now in my environment, need to run a few more jobs, but
> it seems fine here.
What I was actually thinking of doing to correct this is to not set
the next-server line if netboot-enabled=false. So basically we'd
revert Nishanth's change and then modify the dhcpd template:
#if $iface.netboot_enabled:
next-server $next_server;
#end if
This is already being set in modules/manage_isc.py:
interface["netboot_enabled"] = blended_system["netboot_enabled"]
So it should be a nice easy fix. Regardless, whatever we do I'll get
this pushed out as 2.2.3-2, as I didn't think through the impact this
might have to users that manage dhcpd with cobbler.
_______________________________________________
cobbler mailing list
cobbler(a)lists.fedorahosted.org
https://fedorahosted.org/mailman/listinfo/cobbler
5:05 p.m.
On 06/08/2012 12:42 PM, Dan Rao wrote:
Hi all,
I upgraded my Cobbler install to 2.2.3-1.el6 (from epel-testing) last
night.
Interestingly, when I run cobbler sync, only machines that have netboot
enabled get entries in my dhcpd.conf.
The offending line(s) appear to be in
/usr/lib/python2.6/site-packages/cobbler/modules/manage_isc.py - lines
164-165
----
if not interface["netboot_enabled"]:
continue
----
(This is in the section that generates the ISC dhcpd config file).
I commented those lines out to get my server back, but I was curious as
to why that was there in the first place.
Running current git master, which has:
if not interface["netboot_enabled"] and interface['static']:
continue
So, I'm still seeing a problem with this in that in order to netboot machines
with static interfaces I need to run "cobbler sync" after checking the netboot
enable box. This is a real pain.
--
Orion Poplawski
Technical Manager 303-415-9701 x222
NWRA, Boulder Office FAX: 303-415-9702
3380 Mitchell Lane orion(a)nwra.com
Boulder, CO 80301 http://www.nwra.com
5:41 p.m.
So, I'm still seeing a problem with this in that in order to
netboot
machines with static interfaces I need to run "cobbler sync" after checking
the netboot enable box. This is a real pain.
The original issue didn't address that. To overcome that issue, we
need some sort of OMAPI implementation to do dynamic DHCP updates. I
have one solution in mind, but the code is GPL3, so I'm not sure what
implications that'd have on the code base.
We had a discussion on that a while back on IRC, and the consensus was
that if I included GPL3+ code it would effectively bump the GPL2+
license that cobbler currently ships with to GPL3+, which would be an
issue for some contributors who's parent companies have stated they
would not use GPL3.
If anyone has further insight into that, I'd love to hear about it, or
if they know of a native python OMAPI implementation that doesn't just
shell out to the ISC omshell command (what cobbler used to do
pre-2.0).
4286
days inactive
4345
days old
cobbler@lists.fedorahosted.org
23 comments
6 participants
participants (6)
-
Dan Rao -
Detlev Habicht -
Greg Swift -
James Cammarata -
Nishanth Aravamudan -
Orion Poplawski