samgrid installation

The new installation area should be ~sam/jim_products, everything (with a few exception when root is needed) can be installed as user sam. The samgrid installation replace the installation of the JIM products (with exception of the server_run product and the jim_client) and the tailoring of those products. Then a SAM installation is needed and described on this webpage. What is not described here by now are the authentification isssues. This will be done soon.

What is described here is a execution-monitoring-submission-client installation. The pieces of the execution and submission and client installation are fairly well separated already in this instructions, so that any configuration of installation can be put together by the user.

instruction for a client site installation: at the moment only instructions working with sam v5.0.

make ups/upd database for JIM products:

Check that you do not have set up any other environment. The variable $PRODUCTS should be empty.
create a ups/upd database:
- download scripts
- adapt dbconfig.private
- ```
bootups private
```
cd "your directory"/products/setups/
source setups.sh
setup upd

install samgrid and configure it:

upd install -Gc samgrid
ups declare -c samgrid v4_0_3
ups declare -c python v2_1
mkdir /home/sam/jim_samgrid
setup samgrid

This leads to the tailoring dialogue:

samgrid needs to be tailored. Enter the path to JIM area on your machine. This where all log filesand sandboxes will be created: (default /var/local/jim)	/home/sam/jim_samgrid
Enter the experiment for which you are installing the products. If you are installing for multiple experiments you may have to manually retailor some packages (d0 or cdf)	cdf

mkdir upsdb/tomcat
samgrid install config (installs the products)

samgrid install globus

QUERY: Do you want to create default config for sam_gsi_config. This will create a directory structure with all the files under /home_scotgrid/s/sam/jim_samgrid. If you want to use an exisiting sam_gsi_config from some other area answer 'n' and do the subsequent tailoring appropriately (ans: y/n) [default y]	n
Where do you want to store the sam service certificates ( this MUST be a non-exported area writable by user sam) [default: /home_scotgrid/s/sam/jim_samgrid/gsi]?	/home_scotgrid/s/sam/private/gsi
Do you want to configure GSI for gridftp [default: yes]?	default
Are you installing gridftp for d0 or cdf?	cdf
Do you want to configure GSI for the JIM client [default: no]?	yes
Do you want to configure GSI to run a gatekeeper (answer yes if you are installing JIM execution site; you will need access to root) [default: no]?	yes
Do you want to configure GSI for jim_advertise (answer yes if you are also installing a gatekeeper, unless you are an expert) [default: yes]?	yes
Do you want to configure GSI to run a JIM submission site [default: no]?	yes
Do you want to configure GSI to run a JIM gridftp server for intra-cluster transfers [default: no]?	yes

as user root:

 source ~sam/samgrid_inst/products/setups/setups.sh 
          setup samgrid
          setup vdt
          globusPostInstallRoot

(probably you have to run this twice)

ups tailor vdt
setup vdt
sam_cert_request --name=Valeria_Bartsch --email=bartsch@fnal.gov --phone=+44-1865-283111

samgrid install execution_site (I show you an example of the tailoring dialogue here:

1st Step - We will create a site configuration by running "ups configure_site jim_config" 2nd step - jim execution site packages (listed above) will be installed. 3rd step - Basic checks INFO: Lets tailor product jim_config INFO: Executing command: ups configure_site jim_config
Enter the name of the site you want to configure. A single site can comprise DZero and CDF installations. What is the value of 'name' of 'site' ? [FNAL]:	oxford2
Enter the name and architecture of each cluster in your site. Examples of names are clued0 CAB SamGrid Gridka etc. Examples of architectures are Linux+2.4 IRIX+6.5 SunOS+5 OSF1+V4 etc. What is the value of 'architecture' of 'cluster' ? [Linux+2.4]: The value of 'architecture' is set to 'Linux+2.4'
Does this cluster run a gatekeeper? If yes enter the Fully Qualified domain name and the port of each gatekeeper in the format shown in the default value (see later). If this cluster has no gatekeeper but only stations answer 'no' Do you want to configure gatekeeper ? : [no]	yes
What is the value of 'location' of 'gatekeeper' ? [matrix.physics.ox.ac.uk:2119]:	default
Enter the name of each jobmanager configured to run with this gatekeeper. What is the value of 'name' of 'jobmanager' ? [jobmanager-samgrid]:	default
Enter the station name universe and experiment for the stations that are managed by this gatekeeper What is the value of 'name' of 'station' ? [fnal_farm]:	cdf-oxford
What is the value of 'experiment' of 'station' ? [d0]:	cdf
What is the value of 'universe' of 'station' ? [prd]:	default
Enter the station name universe and experiment for the stations that are managed by this gatekeeper Do you want to configure another station ? : [no]	default
Enter the name of each jobmanager configured to run with this gatekeeper. Do you want to configure another jobmanager ? : [no]	default
Does this cluster run a gatekeeper? If yes enter the Fully Qualified domain name and the port of each gatekeeper in the format shown in the default value (see later). If this cluster has no gatekeeper but only stations answer 'no' Do you want to configure another gatekeeper ? : [no]	default
Enter the station name universe and experiment for the stations that cannot be reached via any of the gatekeepers configured above. Do you want to configure station ? : [no]	default
Enter the name and architecture of each cluster in your site. Examples of names are clued0 CAB SamGrid Gridka etc. Examples of architectures are Linux+2.4 IRIX+6.5 SunOS+5 OSF1+V4 etc. Do you want to configure another cluster ? : [no]	default
Enter the location (node and path) where files can be temporarily stored. This location can be used, for example, by runjob to store temporary thumbnails that need to be merged. Do you want to configure local_storage ? : [no]	yes
What is the value of 'path' of 'local_storage' ? [/sam/disk]:	/data5/scratch/sam
What is the value of 'node' of 'local_storage' ? [matrix.physics.ox.ac.uk]:	default

--> latest tailoring dialogue:

INFO: Lets tailor product jim_config INFO: Executing command: ups configure_site jim_config		----------------------- site ----------------------- Enter the name of the site that you are configuring. A single site can comprise DZero and CDF installations. This name is used for monitoring purposes only and you don't need to remember it for other parts of the configuration process. What is the value of 'name' of 'site' ? [FNAL]:	glasgow2
----------------------- cluster ----------------------- Enter the name and architecture of each cluster in your site. The cluster name is used for monitoring purposes only and you don't need to remember it for other parts of the configuration process. Examples of names are clued0, CAB, SamGrid, Gridka, etc. Examples of architectures are Linux+2.4, IRIX+6.5, SunOS+5, OSF1+V4, etc. What is the value of 'architecture' of 'cluster' ? [Linux+2.4]:
What is the value of 'name' of 'cluster' ? [SamGrid]:
----------------------- gatekeeper ----------------------- Enter the fully qualified domain name and the port of your gatekeeper in the format shown in the default value. The gatekeeper runs on the machine where you are installing the execution site software suite. What is the value of 'location' of 'gatekeeper' ? [cdfg.ph.gla.ac.uk:2119]:
----------------------- jobmanager ----------------------- Enter the name of the jobmanager configured to run with this gatekeeper. Accept the default, unless you need to interface grid jobs to mc_farm. What is the value of 'name' of 'jobmanager' ? [jobmanager-samgrid]:
----------------------- station ----------------------- Enter the sam station name universe and experiment for the station at your site What is the value of 'universe' of 'station' ? [prd]:
What is the value of 'name' of 'station' ? [fnal-farm]:	cdf-glasgow
What is the value of 'experiment' of 'station' ? [d0]:	cdf
----------------------- local_storage ----------------------- Enter the location (node and path separately) where files can be temporarily stored. This location will be used, for example, by runjob to store temporary thumbnails that need to be merged. What is the value of 'path' of 'local_storage' ? [/sam/disk]:	/data5/scratch/sam
What is the value of 'node' of 'local_storage' ? [cdfg.ph.gla.ac.uk]:

samgrid install submission_site (giving tailoring dialogue below):
ups declare -c galax 0_2

left-overs after the samgrid product installation

(I have of course already done some things as root. E.g. in the directory /etc/grid-security. So these things are not accounted for in this installation. But this still has to be done at a fresh installation.)

upd install -Gc server_run
ups declare -c server_run v1_2
printf "/home/sam/jim/server_run\nbartsch@fnal.gov\n\n\nyes\n\n\nyes\n\n\nyes\n\n\nyes\n\n\n\nyes\n\n\nyes\n\n\n\n\n" | ups tailor server_run
ups run server_run &
check with:
```
ps -ef | grep sam 
```
and with the monitoring webpages that this suceeded
upd install sam v5_0
ups declare -c sam v5_0
upd install sam_user_api v5_1_0_5
ups declare -c sam_user_api v5_1_0_5
upd install sam_config v4_2_32
ups declare -c sam_config v4_2_32
upd install sam_common v5_1_1_4
ups declare -c sam_common v5_1_1_4
upd install samgrid_util v1_01
ups declare -c samgrid_util v1_01

ups tailor sam_config
and that is the tailoring dialogue:

add configuration	a
3: CDF Default Production Station Installation (__cdf_station_default_prd__)	3
Enter new configuration qualifier:	cdf_prd
Enter new configuration description:	CDF prd
e: edit configuration	e
1: CDF prd (cdf_prd)	1
a: add variable	a
Enter variable to add:	SAM_STATION
Enter value for SAM_STATION:	cdf-oxford
d: done (default)	d
Are you sure that you want to keep these changes?	y
d: done (default)	d

Now need to configure sam_batch_adapter, see http://home.fnal.gov/~bartsch/sam_batch_adapter_docu/

For installing jim_client you need to install sam_gsi_config in a version without the vdt qualifier:

upd install -Gc kx509
         ups declare -c kx509 v2.0
         upd install -Gc sam_gsi_config
         ups declare -c sam_gsi_config v2_2_3

and then tailor sam_gsi_config:

printf "\n\ncdf\nyes\nyes\n\nyes\nyes\n" | ups tailor sam_gsi_config

and the same for the sam_gsi_config with vdt qualifier:

ups tailor sam_gsi_config -q vdt

ups declare -c samgrid_logger_client v0_3
upd install -Gc jim_client v2_0_8
ups declare -c jim_client v2_0_8
printf "physics.ox.ac.uk\n\n\n\nyes\n\n" | ups tailor jim_client
upd install -Gc sam_gridftp
All CDF sites configured with init_sam have the problem that a .k5login file is automatically produced in /home/sam/.k5login. This file has to vanish, otherwise problems with the pbs batch system occur.
A problem for JIM is different time information on the worker nodes and the head node. Therefore it is important to synchronize those. This can be done as user root e.g. with the command:
```
/sbin/service ntpd restart
```

things I did not like about the samgrid installation

whenever run "globusPostInstallRoot" the permissions of the directories: "/etc/grid-security/certificates", "/etc/grid-security/certificates/doegrids" and "/etc/grid-security/certificates/doesciencegrid" are changed and I therfore get a authentication error.
server_run not installed and tailored
sam_gridftp is not installed.

Valeria Bartsch (CDF/GRID)

Last modified: Sun Nov 7 15:50:07 GMT 2004