SAMGRID FAQ’s

Questions

  1. What is the valid format of entries in condor’s grid mapfile at jim_broker, jim_broker_client?
  2. What entries should the condor’s grid-mapfile contain at jim_client?
  3. What entries should the condor’s grid-mapfile contain at jim_broker_client?
  4. What entries should the condor’s grid-mapfile contain at jim_broker?
  5. Which wildcards can be used in condor’s grid-mapfile?
  6. How do I allow anyone to write classads to my system?
  7. What’s the significance of CONDOR_IDS in the run script?
  8. What’s the significance of GSI_DAEMON_NAME in the condor config of jim_broker, jim_broker_client and jim_advertise?
  9. How do I know if my scheduling site is correctly configured?
  10. How do I know if my execution site is correctly configured and running?
  11. I submit the job with a valid proxy but the job does not execute and goes on "HOLD"?
  12. How do I know if my xmldb_server is correctly configured and running?
  13. How do I make jim_broker, jim_broker_client, jim_client, jim_advertise use a particular network interface on my system?
  14. I am not able to able to send any classAd to the broker nor am I able to run the condor_q, condor_status commands to retrieve information from the broker. I have more than one network interface could this be a problem?

 

What is the valid format of entries in condor’s grid mapfile at jim_broker, jim_broker_client?
Subjects of the certificates/proxies enclosed in double quotes followed by the username@domainname which it should be mapped to. First example below is from the sam service certificate and the second example is from the user proxy. Please note that for the user proxy the entry in the gridmap should not contain the “CN=proxy” bit.
Example:
“/C=US/ST=Illinous/L=Batavia/O=Ferminlab/CN=sam/samgrid.fnal.gov” sam@isdsrv1.fnal.gov
“/DC=gov/DC=fnal/O=Fermilab/OU=People/CN=XYZ/USERID=xyz” xyz@fnal.gov

[TOP]

What entries should the condor’s grid-mapfile contain at jim_client?

Condor’s grid mapfile is not required by the jim_client package.

[TOP]

What entries should the condor’s grid-mapfile contain at jim_broker_client?
a. Subject of the jim_broker_client daemons
b. Subject of the jim_broker daemons
c. Subject of all the users who will submit the job to this submission site.
All the subjects should be mapped to a local identity like “user@domain”.

[TOP]

What entries should the condor’s grid-mapfile contain at jim_broker?
a. Subject of the jim_broker_client daemons
b. Subject of the jim_broker daemons
c. Subject of the jim_advertise daemons

[TOP]

Which wildcards can be used in condor’s grid-mapfile?
To accept all the users with valid proxies enter “^.*$” instead of the individual subjects. Note that this is not recommended.

[TOP]

How do I allow anyone to write classads to my system?
To do this you have to disable the security polices used by condor for writing classads. Insert or modify the following configuration parameters in the condor_config file.

SEC_DEFAULT_AUTHENTICATION = NEVER
SEC_READ_AUTHENTICATION = NEVER
SEC_WRITE_AUTHENTICATION = NEVER
SEC_ADMIN_AUTHENTICATION = ALWAYS
SEC_DEFAULT_ENCRYPTION = OPTIONAL
SEC_DEFAULT_INTEGRITY = OPTIONAL

[TOP]

What’s the significance of CONDOR_IDS in the run script?
It tells condor the Linux identity of the user under which the condor daemons are running.

[TOP]

What’s the significance of GSI_DAEMON_NAME in the condor config of jim_broker, jim_broker_client and jim_advertise?
It tells the condor the identity of the server daemons I should expect when I send classads. Setting it to * tells condor to trust all the daemons. This variable could be a comma separated list.

[TOP]

How do I know if my scheduling site is correctly configured and running?
a. Run the command "setup jim_broker_client"
b. Run the command "condor_status -schedd". If your scheduler is properly registered with the broker you should be able to see it.

[TOP]

How do I know if my execution site is correctly configured?
a. Run the command "setup jim_client"
b. Run the command "samg list resources". If your execution site is properly registered with the broker you should be able to see it.

[TOP]

I submit the job with a valid proxy but the job does not execute and goes on "HOLD"?
There could be several reasons for that.
a. Execution site is not configured properly.
b. You are not authorized to run the job at the particular execution site. i.e. your subject is not present in the /etc/grid-security/grid-mapfile.
c. Your proxy life time was short and expired before your job could be submitted to the execution site.

[TOP]

How do I know if my xmldb_server is correctly configured and running?
a. Go to the URL at which the Xindice should be running. it should look something like http://machinename.domain:7080/Xindice. If you have configured tomcat to run at port number other than 7080 (default), use that port number instead of 7080.
Example: http://samgrid.fnal.gov:7080/Xindice

[TOP]

How do I make jim_broker, jim_broker_client, jim_client, jim_advertise use a particular network interface on my system
If you have more than one network interface you need to instruct above mentioned packages to use a specific network interface. This is typical case on systems which have a separate network interface for local network separate from the outside network. Note to use JIM packages you have to use a interface that is available from/to outside world.

Add the following line to the condor_config of these packages -
NETWORK_INTERFACE = <IP address>

To find the IP address do -
$ /sbin/ifconfig

The response to this command should be something like -

eth0              Link encap:Ethernet HWaddr 00:A0:C9:D4:39:CE
                    inet addr:131.225.224.195 Bcast:131.225.227.255 Mask:255.255.252.0
                    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
                    RX packets:7370338 errors:0 dropped:0 overruns:0 frame:0
                    TX packets:3221810 errors:0 dropped:0 overruns:0 carrier:0
                    collisions:0 txqueuelen:100
                    Interrupt:5 Base address:0x6000

eth1              Link encap:Ethernet HWaddr 00:E0:18:1C:4A:75
                    inet addr:192.168.1.2 Bcast:192.168.1.255 Mask:255.255.255.0
                    inet6 addr: fe80::2e0:18ff:fe1c:4a75/10 Scope:Link
                    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
                    RX packets:1533557 errors:0 dropped:0 overruns:0 frame:0
                    TX packets:935306 errors:0 dropped:0 overruns:0 carrier:0
                    collisions:0 txqueuelen:100
                    RX bytes:213577142 (203.6 Mb) TX bytes:271993773 (259.3 Mb)
                    Interrupt:10 Base address:0x8000

lo                  Link encap:Local Loopback
                    inet addr:127.0.0.1 Mask:255.0.0.0
                    UP LOOPBACK RUNNING MTU:16436 Metric:1
                    RX packets:485461 errors:0 dropped:0 overruns:0 frame:0
                    TX packets:485461 errors:0 dropped:0 overruns:0 carrier:0
                    collisions:0 txqueuelen:0

Finding the IP address that is not a private address and substitute it in the above configuration line.

[TOP]

I am not able to able to send any classAd to the broker nor am I able to run the condor_q, condor_status commands to retrieve information from the broker. I have more than one network interface could this be a problem?
This could possibly be a problem. See FAQ 13 for resolution.

[TOP]


Email your questions/suggestions to d0-gritt@fnal.gov