On 27/08/2014 01:06, Reuti wrote:
> Hi,
>
> Am 27.08.2014 um 00:53 schrieb Peter van Heusden:
>
>> I'm currently installing a worker node on our cluster and using Ubuntu
>> 14.04 as the operating system of the worker node. I've installed the
>> gridengine-exec package, which is version 6.2u5-7.3. When I add the exec
>> node, however, the qmaster complains that:
>>
>> got load report from host "ehost.mydomain.org" - reports load value for
>> host "ehost"
>>
>> The host itself is configured with the hostname ehost.mydomain.org, and
>> it has the same FQDN in the host entry in qconf. The bootstrap file states:
>>
>> default_domain          none
>> ignore_fqdn             false
>>
>> If I watch the traffic on port 6444, the new exec indeed reports as
>> ehost, not ehost.mydomain.org. If I look at the same traffic on one of
>> our older (Ubuntu 12.04) hosts, they report as their FQDN. As a result
>> of this (I think) qhost shows no info from the new execution hosts and
>> jobs cannot run on that host.
>>
>> Does anyone know how to make this new host report as its FQDN?
> How is the name resolved - /etc/hosts or DNS? Has it more than one network 
> interface?
Thanks, this gave me the clue of where to look. Host resolution is via
DNS, but the host had a bare hostname in /etc/hosts, so:

127.0.1.1       ehost

I changed that to

127.0.1.1       ehost.mydomain.org ehost

and restarted gridengine-exec and now its working!


> There are tools in "$SGE_ROOT/utilbin" to investigate this like 
> `gethostbyaddr`...
>
> -- Reuti
>
>
>> Thanks!
>> Peter
>>
>>
>>
>> _______________________________________________
>> users mailing list
>> users@gridengine.org
>> https://gridengine.org/mailman/listinfo/users

_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users

Reply via email to