[gold-users] Gold no longer working

Scott Jackson scottmo at adaptivecomputing.com
Sun Jul 29 21:27:45 MDT 2012


On Sun, Jul 29, 2012 at 5:10 PM, Shade Alabsa <shade34321 at gmail.com> wrote:

> So I recently upgraded my gold installation to Gold 2.2.0.5 and it's still
> not workingBelow are the steps I took to upgrade it, please let me know if
> I missed any steps and/or made any mistakes in my configuration. I did not
> make the gold user since previously we ran all gold commands as root and I
> decided to keep it the same. I went into the gold-2.2.0.4 direcotry and ran
> the configure script with
>     ./configure --prefix=/usr/local/ --with-db=Pg
>     make
>     make deps
>     make install
>     make auth_key
>

In these steps, ./configure should be run by the (preferably non=root) gold
admin user, while make deps, make install and make auth_key should be run
as root. Did you do that?


> It was previously installed under /usr/local so I again I kept it the same
> and I entered my secret key when prompted, to make sure I didn't mess that
> up we'll say I used SSS.
>
>
I then edited /usr/local/gold/etc/goldd.conf and
> /usr/local/gold/etc/gold.conf and edited them to look similar to what the
> previous ones looked like. I essentially changed server.host to the host
> name, mgt.cluster.net,  and database.datasource host part to reflect the
> hostname as well. I then started to configure the database, since I was
> using an existing database I didn't really do much in this section but to
> note I wasn't able to edit /etc/postgresql/../.../pg_hba.conf since one
> didn't exist there. I did a locate on the file and found it was located at
> /var_lib_new/pgsql/data/pg_hba.conf and I added this line
> host all all 172.20.0.1 255.255.255.255 trust
>
> 172.20.0.1 is the internal IP address for our management node. I had the
> same problem with postgresql.conf as well but this is the line I added to it
>
> listen_addresses = 'mygoldserver,mgt.cluster.net'. Now I skipped the add
> gold user as a database administrator and create the gold database since I
> already had a database I wanted to use. I also skipped the web server setup
> since we don't have a webserver set up.
>
> Now to tie Gold into the system I edited /usr/local/maui/maui.cfg and
> added this line
>
> AMCFG[bank] TYPE=GOLD HOST=mgt.cluster.net SOCKETPROTOCOL=HTTP
> WIREPROTOCOL=XML CHARGEPOLICY=DEBITALLWC JOBFAILUREACTION=IGNORE TIMEOUT=15
>
> I took out the PORT part since it was commented out in the previous parts.
> I also edited /usr/local/maui/maui-private.cfg and added this line, it was
> empty before,
>
> CLIENTCFG[AM:bank] KEY=SSS AUTHTYPE=HMAC64
>
> That all sounds about right.


> This was getting gold to give this error message when I tried to use gusage
>
> Aborting account statement: Failed sending message: (Unable to connect to
> socket (Connection timed out)).
>
> This is a different message than before. Before you were getting a problem
with authorization, now the client apparently cannot locate a running
server.
This could happen if the goldd daemon was not running, or if the server is
listening on a different port than the clients are trying to speak on.

Do the server and ports match up in the gold.conf and goldd.conf?
Is goldd running? Are both clients and servers being started up under the
/usr/local path?




> Now I'm getting this error
>
>  DBI connect('dbname=gold;host=mgt.cluster.net','root',...) failed:
> FATAL:  role "root" does  not exist at /usr/local/gold/lib/Gold/Database.pm
> line 165
>
>
Hmmm, yet a different error. Did you resolve the Unable to connect to
socket error? For thus it would appear.

This might suggest that previously your SystemAdmin Role was associated
with a non-root user and your database would only recognize the old Gold
admin user as having admin authority. Perhaps you ran configure this time
as root, where it had been run as non-root before? Did you perhaps try to
start goldd as root now, whereas you were starting it as non-root before?





> Any help you can offer would be great! Thanks!!
>
>
> Shade
>
> On Fri, Jul 27, 2012 at 5:16 PM, Shade Alabsa <shade34321 at gmail.com>wrote:
>
>> I looked into what you said and it looks like GOLD_HOME is unset among a
>> few other things. Part of the reason we updated everything was because we
>> had last /etc, long story, so of course we don't have a /etc/gold.cfg. As
>> of right now I think I'm going to update it as well and go from there.
>> Thanks for the help!
>>
>> Shade
>>
>>
>> On Thu, Jul 26, 2012 at 7:21 PM, Scott Jackson <
>> scottmo at adaptivecomputing.com> wrote:
>>
>>> What is your path to glsuser?
>>>
>>> `which glsuser`
>>>
>>> Is it /usr/local/gold/bin/glsuser or might there be another installation
>>> that got partially installed?
>>>
>>> Also, do you have GOLD_HOME in your environment.
>>>
>>> `env | grep GOLD_HOME`
>>>
>>> Perhaps this has been set or unset from what it was before.
>>>
>>> Is there an /etc/gold.cfg?
>>>
>>> What is the use lib path at the top of your glsuser file?
>>>
>>> `grep 'use lib' `which glsuser``
>>>
>>> use lib qw(/opt/gold-upgrade/lib /opt/gold-upgrade/lib/perl5);
>>>
>>> This will tell you where the Gold.pm is that it is using.
>>>
>>> From the Gold.pm, it builds the location ($PREFIX)  for the auth_key
>>> file.
>>>
>>>
>>>    # Set $PREFIX from Environment variable, Cfg File, or Hard-coded
>>> default
>>>
>>>     if ($ENV{GOLD_HOME}) { $PREFIX = $ENV{GOLD_HOME}; }
>>>     elsif (-e "/etc/gold.cfg")
>>>     {
>>>         my ($var, $val);
>>>         open(ENV, "/etc/gold.cfg");
>>>         while (<ENV>)
>>>         {
>>>             chomp;       # Remove trailing whitespace
>>>             s/^\s+//;    # Remove leading whitespace
>>>             s/\#.*//;    # Strip off comments
>>>             next unless ($var, $val) = split("[= ]+", $_, 2);
>>>             $ENV{$var} = $val;
>>>         }
>>>         $PREFIX = $ENV{GOLD_HOME};
>>>     }
>>>     else { $PREFIX = "/opt/gold"; }
>>>
>>>    open AUTH_KEY, "${PREFIX}/etc/auth_key"
>>>
>>>
>>>
>>> On Thu, Jul 26, 2012 at 4:07 PM, Shade Alabsa <shade34321 at gmail.com>wrote:
>>>
>>>> Ideally I'd like to upgrade Gold but at the time we were only doing
>>>> what was necessary to get our cluster back up and running.
>>>>
>>>> Gold was installed to /usr/local/ so I checked there and there still is
>>>> a auth_key file,
>>>>
>>>> [root at mgt maui]# locate auth_key
>>>> /usr/local/gold/etc/auth_key
>>>> /usr/local/gold-2.1.6.0/etc/auth_key
>>>> /usr/local/src/gold-2.1.11.0/etc/auth_key
>>>>
>>>> I'm assuming gold-2.1.11.0 is the old installation. I'm not sure what
>>>> you mean by resync it with CSKEY value in maui-private.cfg, mainly becasue
>>>> my maui-private.cfg file has nothing in it. Thanks for the speedy response
>>>> and sorry for the long delay.
>>>>
>>>> Shade
>>>>
>>>> On Thu, Jul 26, 2012 at 2:07 PM, Scott Jackson <
>>>> scottmo at adaptivecomputing.com> wrote:
>>>>
>>>>> So, you didn't upgrade Gold?
>>>>>
>>>>> It appears you may have deleted/moved the /opt/gold/etc/auth_key file.
>>>>> You should be able to regenerate it by running make auth_key as root from
>>>>> your original gold tarball unpack directory. You will have to use the same
>>>>> key as before or else resync it with the CSKEY value in maui-private.cfg.
>>>>>
>>>>> I hope this helps.
>>>>>
>>>>> Scott
>>>>>
>>>>>
>>>>> On Thu, Jul 26, 2012 at 11:46 AM, Shade Alabsa <shade34321 at gmail.com>wrote:
>>>>>
>>>>>> We are currently using the Gold accounting manager, version 2.1.6.0.
>>>>>> Recently we had to upgrade our xcat, torque, and maui installations and now
>>>>>> Gold is not working anymore. I get the following errors:
>>>>>>
>>>>>> [root at mgt ~]# glsuser
>>>>>> Unable to open auth_key file: No such file or directoryBEGIN
>>>>>> failed--compilation aborted at /usr/local//lib/Gold.pm line 150.
>>>>>> Compilation failed in require at /usr/local/bin/glsuser line 57.
>>>>>> BEGIN failed--compilation aborted at /usr/local/bin/glsuser line 57.
>>>>>> [root at mgt ~]# gusage -s 2012-07-01 -e 2012-07-31 -p CCL >>
>>>>>> garuda-summary-July_12.log
>>>>>> Unable to open auth_key file: No such file or directoryBEGIN
>>>>>> failed--compilation aborted at /usr/local//lib/Gold.pm line 150.
>>>>>> Compilation failed in require at /usr/local/bin/gusage line 53.
>>>>>> BEGIN failed--compilation aborted at /usr/local/bin/gusage line 53.
>>>>>>
>>>>>> I'm not quite sure what the problem is or if our upgrades broke it.
>>>>>> Any help you can provide would be great! Thanks!
>>>>>>
>>>>>> Shade
>>>>>>
>>>>>> _______________________________________________
>>>>>> gold-users mailing list
>>>>>> gold-users at supercluster.org
>>>>>> http://www.supercluster.org/mailman/listinfo/gold-users
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> gold-users mailing list
>>>>> gold-users at supercluster.org
>>>>> http://www.supercluster.org/mailman/listinfo/gold-users
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> gold-users mailing list
>>>> gold-users at supercluster.org
>>>> http://www.supercluster.org/mailman/listinfo/gold-users
>>>>
>>>>
>>>
>>> _______________________________________________
>>> gold-users mailing list
>>> gold-users at supercluster.org
>>> http://www.supercluster.org/mailman/listinfo/gold-users
>>>
>>>
>>
>
> _______________________________________________
> gold-users mailing list
> gold-users at supercluster.org
> http://www.supercluster.org/mailman/listinfo/gold-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.supercluster.org/pipermail/gold-users/attachments/20120729/70db746c/attachment-0001.html 


More information about the gold-users mailing list