Re: [Condor-users] condor_master dying in 7.6.9

[Roderick Johnstone <rmj@xxxxxxxxxxxxx>]

Brian

I've been travelling and had hoped to get back to you while I was away
but unfortunately that turned out to be difficult...

On 18/08/12 18:05, Brian Candler wrote:
> On Sat, Aug 18, 2012 at 05:45:35PM +0100, Roderick Johnstone wrote:
>> $ grep -i localhost /etc/hosts
>> 127.0.0.1       localhost       localhost.localdomain   localhost4
>> localhost4.localdomain4
>> ::1     localhost       localhost.localdomain   localhost6
>> localhost6.localdomain6
>> 127.0.0.1       localhost.localdomain   localhost
>> ::1     localhost6.localdomain6 localhost6
> 
> So firstly you have two lines each for 127.0.0.1 and ::1; and secondly,
> "localhsot" and "localhost.localdomain" is on both.
> 
> 127.0.0.1	localhost.localdomain localhost localhost4.localdomain localkhost4
> ::1		localhost6.localdomain localhost6
> 

So, I removed the second lines for 127.0.0.1 and ::1 and it doesnt seem
to help. For those addresses I now have:

127.0.0.1       localhost       localhost.localdomain   localhost4
localhost4.localdomain4
::1     localhost       localhost.localdomain   localhost6
localhost6.localdomain6


> But it's odd anyway that Condor is trying to use these. Possibly the
> hostname in /etc/hostname is wrong.
> 
> You should have /etc/hostname containing a fully-qualified domain name, e.g.:
> 
> pc1.mydomain.com
> 
> After changing this, do
> 
> hostname $(cat /etc/hostname)

I don't have a /etc/hostname file. On redhat that seems to be set in
/etc/sysconfig/network where it seems to be set correctly since the
hostname command returns the correct FQDN of the host.

> 
> Then you should ensure you have an entry in /etc/hosts with your interface
> IP:
> 
> x.x.x.x  pc1.mydomain.com pc1

That seems to be set correctly too.

> 
> Unless you already have external DNS set up properly, with forward DNS
> mappping pc1.domain.com to x.x.x.x, and reverse DNS mapping x.x.x.x to
> pc1.mydomain.com, in which case the hosts entry is not required.

Running nslookup on the FQDN returns the correct ip address and running
nslookup on the ip address returns:

xxx.xxx.xxx.xxx.in-addr.arpa      name = a.b.c.d

where xxx.xxx.xxx.xxx is now the ip address in reverse order and a.b.c.d
is the FQDN.

I also have the strace output from running condor_master that you
suggested I gather in another email to this list. Would you be prepared
to have a look at that off list? If so, would it be ok to email that to
you. Its about 650kB.

Thanks again.

Roderick

> 
> Regards,
> 
> Brian.
>