On Thu, Feb 13, 2014 at 11:05 AM, SCHAER Frederic
wrote:
> Hi,
>
> Thanks for the suggestion. I feared something like that might be suggested
> though ;)
> I'll try to retrieve 6.5 and try to validate it before the affected servers
> go live.
>
> Please note we're not using NM : it is on of the the first things we disable
> when setting up the nodes after the initial reboot, but anaconda seems to be
> using it and unless I'm mistaken we don't have control on anaconda's
> preferred way of handling the network, do we ?
Not without pullint out one of *my* favorite tricks! Build a clean OS
image tarball, and use any live CD or kickstart or any other PXE based
tool you wish to partitioni the target disk, drop the tarball onto it,
tweak a few config settings as needed, run "grub-install" from inside
the new installation's chroot cage, and reboot.
It takes work to tun, but it is *much, much, much* faster than pulling
down RPM's and assembling the new environment from scratch. And
without the bulky and discarded "yum" metadata downloads it's much
more network efficient..
>
> Regards
>
> De : Nico Kadel-Garcia [nka...@gmail.com]
> Envoyé : jeudi 13 février 2014 03:55
> À : SCHAER Frederic
> Cc : SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV
> Objet : Re: SL6.2 randomly stays idle during install
>
> Install 6.5? There are significant library and kernel updates since SL
> 6.5, especially for the latest and greatest 10G network drivers.
>
> Also, don't use any more of NetworkManager than you have to for servers.
>
> On Wed, Feb 12, 2014 at 10:56 AM, SCHAER Frederic
> wrote:
>> Hi,
>>
>>
>>
>> I am facing a strange issue here : I have installed SL6.2 on several Dell
>> powerdge R720xd, and on some of them, the installs stopped before the
>> reposetup step.
>>
>> I could get a console, play with the logs (I did not save them, I regret
>> that), network was up, but there was absolutely nothing going on.
>>
>>
>>
>> I've read it could be because the networkmanager may think the network is
>> down (while it's not), would fail to bing it up (because it's already up),
>> and that could be what would prevent the install from going on...
>>
>> If I do restart the nodes, there is a chance they will install, and another
>> chance I might need to reboot again.
>>
>>
>>
>> The nodes are booting on a 10G Ethernet card labeled p2p1 (Broadcom
>> Corporation NetXtreme II BCM57810 10 Gigabit Ethernet (rev 10)) , and the
>> em* devices aren't connected.
>>
>> The pxelinux passes this kind of arguments to the kernel :
>> ksdevice=00:0a:f7:24:xx:xx network kssendmac nomodeset
>>
>> The kickstart contains this : network --bootproto dhcp -hostname myhostname
>>
>>
>>
>> Any idea of what could be done to prevent this kind of thing from happening
>> ?
>>
>>
>>
>> Thanks && regards