[GE users] Default master's domain name during installation

Reuti reuti at Staff.Uni-Marburg.DE
Tue Sep 2 11:23:45 BST 2008


Hi,

Am 01.09.2008 um 22:01 schrieb Michael Green:

> Fresh install of 6.1 U5 here.
>
> My master (mcluster05 - shiny new X2200) has two legs:
> 1. Organization network
> 132.77.XXX.XXX/23 - wisdom.someorg.com
>
> 2. Cluster network
> 192.168.1.XXX/24 - mcl.someorg.com
>
> in resolv.conf:
> search mcl.someorg.com wisdom.someorg.com
>
> During the installation of qmaster I choose to vote yes for "all hosts
> of your cluster in a single DNS domain" because all computers of the
> grid are on 192.168.1.X network, in mcl.someorg.com domain.
>
> However for a reason that is beyond my understanding the installer
> decides to tie shadow master to the wisdom.someorg.com domain instead
> of mcl.someorg.com.
> [root at mcluster05 common]# qconf -sh
> mcluster05.mcl.weizmann.ac.il  <-- master
> mcluster05b.wisdom.weizmann.ac.il <-- shadow master (wrong domain)
> n21.mcl.weizmann.ac.il <-- exec
> n22.mcl.weizmann.ac.il <-- exec
> n23.mcl.weizmann.ac.il <-- exec
> n24.mcl.weizmann.ac.il <-- exec

can you try the programs:

gethostbyaddre
gethostbyname

in $SGE_ROOT/utilb/$ARC whether they have the correct response. Has  
the shadow machine two network interfaces, and the primary one is the  
external one (in which case you would need a host_aliases in  
$SGE_ROOT/default/common).

Are the entries in /etc/hosts of your NIS/DNS consistent, and have  
the correct order in /etc/nsswitch.conf?

-- Reuti


> (During the addition of  the execution nodes they are all identified
> correctly as belonging to mcl.someorg.com domain).
>
> [root at mcluster05 lx24-amd64]# ./gethostbyname -all mcluster05
> Hostname: mcluster05.mcl.someorg.com
> SGE name: mcluster05.mcl.someorg.com
> Aliases:  mcluster05
> Host Address(es): 192.168.1.253
>
> root at mcluster05 lx24-amd64]# ./gethostbyaddr -all 192.168.1.253
> Hostname: mcluster05.mcl.someorg.com
> SGE name: mcluster05.mcl.someorg.com
> Aliases:  mcluster05
> Host Address(es): 192.168.1.253
>
> However
> root at mcluster05 lx24-amd64]# ./gethostbyname -all mcluster05b
> Hostname: mcluster05b.wisdom.someorg.com
> SGE name: mcluster05b.wisdom.someorg.com
> Aliases:  mcluster05b
> Host Address(es): 132.77.XXX.YYY
>
> Why?
> Is this because what I have in /etc/hosts:
>
> [root at mcluster05 lx24-amd64]# cat /etc/hosts
> 127.0.0.1       localhost.localdomain   localhost
> 192.168.1.253   mcluster05.mcl.someorg.com   mcluster05
> # 192.168.1.252   mcluster05b.mcl.someorg.com  mcluster05b
>
> I'm pretty sure that I have properly configured BIND serving the
> mcl.someorg.com domain just for the purpose of satisfying Grid
> resolving needs:
> root at mcluster05 lx24-amd64]# host mcluster05
> mcluster05.mcl.weizmann.ac.il has address 192.168.1.253
> [root at mcluster05 lx24-amd64]# host mcluster05b
> mcluster05b.mcl.weizmann.ac.il has address 192.168.1.252
>
> [root at mcluster05 lx24-amd64]# host 192.168.1.253
> 253.1.168.192.in-addr.arpa domain name pointer  
> mcluster05.mcl.someorg.com.
> 253.1.168.192.in-addr.arpa domain name pointer ns.mcl.someorg.com.
> [root at mcluster05 lx24-amd64]# host 192.168.1.252
> 252.1.168.192.in-addr.arpa domain name pointer  
> mcluster05b.mcl.someorg.com.
>
> So, how does the installer decides which of the two (or more) domain
> names available on the master system to use?
>
> -- 
> Warm regards,
> Michael Green
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net


---------------------------------------------------------------------
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