[GE users] small name conflict for execution nodes

Nicolas Joly njoly at pasteur.fr
Wed Mar 2 17:05:59 GMT 2005


On Wed, Mar 02, 2005 at 10:28:33AM -0500, Sean Dilda wrote:
> Nicolas Joly wrote:
> >Hi,
> >
> >I'm trying to set up SGE 6.0u3 on some AMD64 boxes running Linux
> >(CentOS 3.4), and encountered a problem where execution nodes "small
> >names" conflicts ...
> >
> >The problem is that we have execution nodes on 2 domains, but they
> >share the same "small name" :
> >
> >      node1.xx.pasteur.fr	node1.yy.pasteur.fr
> >      node2.xx.pasteur.fr	node2.yy.pasteur.fr
> >      node3.xx.pasteur.fr	node3.yy.pasteur.fr
> >      [...]			[...]
> >
> >The qmaster host (head1.xx.pasteur.fr) was installed without
> >difficulty, with `ignore_fqdn' set to false and `default_domain' to
> >none.
> >
> >I noticed that, at least, SGE uses the "small host name" to generate
> >the execution nodes spool directories ...
> >
> >Is that configuration currently supported ?
> 
> Have you considered using local spool directories for the compute nodes?

Not yet, but will try ...


In the mean time, i want to report an installation problem for
execution hosts (with ignore_fqdn=false and default_domain=none).

The installation script "util/install_modules/inst_execd.sh" always
remove the domain part of the host name; which looks wrong, as it
check for dots a little later to eventually append the default domain,
if exists.

[...]
This hostname is not known at qmaster as an administrative host.

Real hostname of this machine:                     raclette-05.calcul.pasteur.fr
Aliased hostname (if "host_aliases" file is used): raclette-05.calcul.pasteur.fr
Default domain ("none" means ignore):              none
Ignore domain names:                               false

The resulting hostname is:              =========> raclette-05
[...]

njoly at raclette-05 [adm/sge]> ./utilbin/lx24-amd64/gethostname -all
Hostname: raclette-05.calcul.pasteur.fr
SGE name: raclette-05.calcul.pasteur.fr
Aliases:  
Host Address(es): 157.99.69.25 

I checked that the attached patch fix that problem.

Thanks.

-- 
Nicolas Joly

Biological Software and Databanks.
Institut Pasteur, Paris.


    [ Part 2, Text/PLAIN 20 lines. ]
    [ Unable to print this part. ]


    [ Part 3: "Attached Text" ]

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net



More information about the gridengine-users mailing list