Re: [Condor-users] My Pool is empty--why?

["Taylor, Andrew" <Andrew.L.Taylor@xxxxxxxxxxxxx>]

Hi Dave

Looks like a similar problem to what I had.  For some reason * was not working.  I set read and write to xxx.xxx.* and it worked.  What distro are you using?  I had to bin a Ubuntu x64 installation as I just could not get condor_status to show the pool.  

Thanks
 
Andrew Taylor


-----Original Message-----
From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Dave Schneider
Sent: 13 March 2008 21:51
To: Condor-Users Mail List
Subject: [Condor-users] My Pool is empty--why?

Hi,

I've recently installed Condor 7.0.1  using an existing set of local config files from 6.8.x, and I can't seem to get any machines to show up in my queue. For testing purposes, I started up a central manager (which is also an execute node) and one machine that should just be an execute node. Here's what I'm seeing from the central manager and the execute node (running as the condor user):

~/hosts/george/log$ condor_status

~/hosts/george/log$


The only log file that has scary things in it is the CollectorLog, where I see this:


3/13 16:41:46 ******************************************************
3/13 16:41:46 ** condor_collector (CONDOR_COLLECTOR) STARTING UP
3/13 16:41:46 ** /home/condor/condor-7.0.1_64/sbin/condor_collector
3/13 16:41:46 ** $CondorVersion: 7.0.1 Feb 26 2008 BuildID: 76180 $
3/13 16:41:46 ** $CondorPlatform: X86_64-LINUX_RHEL3 $
3/13 16:41:46 ** PID = 12958
3/13 16:41:46 ** Log last touched time unavailable (No such file or 
directory)
3/13 16:41:46 ******************************************************
3/13 16:41:46 Using config source: /home/condor/condor_config
3/13 16:41:46 Using local config sources:
3/13 16:41:46    /home/condor/hosts/vela/condor_config.local
3/13 16:41:46 DaemonCore: Command Socket at <10.1.1.81:9618>
3/13 16:41:46 In ViewServer::Init()
3/13 16:41:46 In CollectorDaemon::Init()
3/13 16:41:46 In ViewServer::Config()
3/13 16:41:46 In CollectorDaemon::Config()
3/13 16:41:46 enable: Creating stats hash table
3/13 16:41:50 (Sending 0 ads in response to query)
3/13 16:41:50 Got QUERY_STARTD_PVT_ADS
3/13 16:41:50 (Sending 0 ads in response to query)
3/13 16:41:50 NegotiatorAd  : Inserting ** "< vela.cyc.com >"
3/13 16:41:50 stats: Inserting new hashent for 
'Negotiator':'vela.cyc.com':'10.1.1.81'
3/13 16:41:54 Got QUERY_STARTD_ADS
3/13 16:41:54 (Sending 0 ads in response to query)
3/13 16:41:54 DaemonCore: PERMISSION DENIED to unknown user from host 
<10.1.1.81:34615> for command 1 (UPDATE_SCHEDD_AD), access level 
ADVERTISE_SCHEDD
3/13 16:41:55 DaemonCore: PERMISSION DENIED to unknown user from host 
<10.1.1.81:34616> for command 2 (UPDATE_MASTER_AD), access level 
ADVERTISE_MASTER
3/13 16:41:57 Got QUERY_STARTD_ADS
3/13 16:41:57 (Sending 0 ads in response to query)
3/13 16:41:59 Got QUERY_STARTD_ADS
3/13 16:41:59 (Sending 0 ads in response to query)
3/13 16:42:02 Got QUERY_STARTD_ADS
3/13 16:42:02 (Sending 0 ads in response to query)
3/13 16:42:03 DaemonCore: PERMISSION DENIED to unknown user from host 
<10.1.1.81:34617> for command 0 (UPDATE_STARTD_AD), access level 
ADVERTISE_STARTD
3/13 16:42:04 DaemonCore: PERMISSION DENIED to unknown user from host 
<10.1.1.81:34618> for command 0 (UPDATE_STARTD_AD), access level 
ADVERTISE_STARTD

10.1.1.81 is the central manager.  Later on there are similar access 
denied message from the execute node.

I tried putting these settings in the master condor_config file (btw, 
even though these are mentioned in the manual for security stuff, 
they're not in the example config files, only the hostallow* setting are 
in there).

ALLOW_READ            = */*
ALLOW_WRITE           = */*
ALLOW_ADMINISTRATOR   = condor@xxxxxxx/*
ALLOW_CONFIG          = condor@xxxxxxx/*
ALLOW_NEGOTIATOR      = condor@xxxxxxx/*
ALLOW_DAEMON          = condor@xxxxxxx/*


Previously, I also tried to use these security settings, but they 
resulted in the same problem:

HOSTALLOW_ADMINISTRATOR = $(CONDOR_HOST)
HOSTALLOW_OWNER = $(FULL_HOSTNAME), $(HOSTALLOW_ADMINISTRATOR)
HOSTALLOW_READ = *
HOSTALLOW_WRITE = *
HOSTALLOW_NEGOTIATOR = $(CONDOR_HOST)
HOSTALLOW_NEGOTIATOR_SCHEDD = $(CONDOR_HOST), $(FLOCK_NEGOTIATOR_HOSTS)
HOSTALLOW_WRITE_COLLECTOR = $(HOSTALLOW_WRITE), $(FLOCK_FROM)
HOSTALLOW_WRITE_STARTD    = $(HOSTALLOW_WRITE), $(FLOCK_FROM)
HOSTALLOW_READ_COLLECTOR  = $(HOSTALLOW_READ), $(FLOCK_FROM)
HOSTALLOW_READ_STARTD     = $(HOSTALLOW_READ), $(FLOCK_FROM)

Anyone have any ideas? I checked DNS (as mentioned in a post to 
condor-users from some time ago), and it appears that the manager is 
findable via both name and number.