Usage of the IT cluster by the Online Software - hints for ONLSW testers
1. The atlonl AFS home
directory
The hone directory basically contains some directories for tests done during
passed scallability tests.
The things of interest for this large scale test period are located in:
public/lst04/
There is a file called ipc_root.ref which contains the IPC reference when
not using a fixed port number for the IPC server, otherwise it can be located
in /pool/online/online-00-21-01/installed/com/ipc_root.ref
Under public/lst04/ there is a directory called tar_files which contains
the tar files of the system installation in /pool/online.
It also contains a few scripts used for installing those tar files, ie
install_db, install_online, and install_update_db.
The is also a directory called gen_info. The following files are all located
in that directory. The list of hosts is located in the following files: batch_hosts_040304_1334.txt,
batch_hosts_040304_1810.txt, hosts_040304_1334.txt, hosts_040304_1810.txt,
machine-info_040304_1810.csv, and machine-info_040304_1810.html. The first
two contain the information as given by the bhosts command (ie unsorted, with
LSF batch info), the two next ones the same info but with just the host names
for the machines that are up (sorted in alphabetical order), the next one
contains the system info for these machines in csv format, and the last one
this information as an HTML table.
There is also a JPG file with the network topology of the machines that
are allocated to us in the gen_config directory. The file name is: atlas-online_network_topology_2004.jpg
An example .onlnerc file is also in the gen_config directory as well as
the SSH config file that people can use to avoid getting asked for accepting
the host key. This file is called ssh_user_config.
The .onlinerc is in the home directory.
The public/lst04 directory should also contains the TDAQ_IPC_INIT_REF
file for an IPC sever running on a fixed machine with fixed port number (when
it has been done).
2. The local disk (/pool/online)
It is organised in the following way:
online-00-21-01
j2sdk1.4.2_03
logs
The release 00-21-01 and the jdk1.4.2 are installed and are setup to work
together.
The logs directory is where most of the logs go to
3. Running the tests
The .cshrc (called from the .tcshrc) sources the release setup script and
sources the get_online_env.csh script.
To run play_daq type:
play_daq <partition> no_obk
All logs go to the /pool/online/logs directory.
The OBK setup is to be checked.
4. WARNINGS
See the electronic log book for details of possible problems and usage of
the cluster by
other people.