Communicating with the Users

Users of the Allegro computing system are informed about important events through messages that are displayed in the console. This chapter gives an overview of how to display and modify the messages.

Note

All kinds of messages are displayed by default when the rc files (see General Setup) are sourced.

Login message

This type of message is supposed inform the user about changes in the system setup or any other important nMessagingews. There is a distinction between news for all users, and news for Allegro members. All news can be displayed by

$ display_welcome_user.sh
$ display_welcome_allegro.sh
=== Info for Allegro members ===
25-Aug-2014: public data projects (aka user-defined projects) are
            renamed to "open projects". All
            aliases change to P_open_* instead of P_up_*
25-Aug-2014: For project management, feel free use the project manager
            wizard. You can start it with "pm_wizard"

The display messages can be modified by Allegro members by typing

$ display_welcome_user.sh --modify

Maintenance

Allegro computers also need maintenance, and an occasional reboot. This reboot is performed by the computer group, which can be contacted through helpdesk@strw.leidenuniv.nl. The user must be informed about the time of the reboot.

Scheduling Maintenance

Maintenance can be scheduled by running

$ schedule_maintenance

This opens a GUI that shows you the current maintenance status.

Clicking on Schedule Maintenance opens another window that allows you to set the date and time of the reboot, the expected downtime, the computers that are affected, and the reason for the reboot. Optionally, the users of the Allegro computers can be informed via email.

The GUI then generates the maintenance file, which is defined by the environment variable $ALLEGRO_MAINTENANCE. If that file exists, a maintenance is scheduled.

If maintenance is scheduled, this will not only be displayed in the GUI, but it can also be seen in the terminal

$ display_maintenance.sh
=== Maintenance ===
computer:  chaxa, tebinquiche, helada, salada, escondida
time:      Thu, 23-Apr-2015, 09:00
downtime:  30 min
reason:    updates

If you have any objections, please contact Allegro.

Warning

Users should be informed at least 3 days in advance about a reboot.

Modifying the Maintenance Date

In case a user objects to the proposed date, you modify the date for maintenance by starting the GUI

$ schedule_maintenance

and click on the Re-Schedule Maintenance button in the GUI. Change the parameters and press OK.

Clearing Mainteance

Once the maintenance operation has finished, the maintenance should be cleared (and users should be informed). This can be done by simply firing up the maintenance-GUI and press the Clear Maintenance button. This simply deletes the maintenance file $ALLEGRO_MAINTENANCE.

Mailing all users

Eventually I will write a little gui thing which allows you to do this, but at the moment the best workaround is to start up an interactive python session on chaxa and get a list of all the addresses as follows.

$ python
>>> from allegroUtils import projectManager
>>> addressList=projectManager.all_users().values()

You will need to remove the single quotes before you can use the list in an email. One way to do this is as follows.

>>> print ','.join(addressList)

Note also that the call to all_users takes some time because it gets the list of users by seeing who has a directory on one of the Allegro file systems.

There are users who may have a directory on the system, but who don’t want to receive mails; and there are users with no project linkage or directory, who do want to receive mails. These people can be catered for by entering their usernames respectively in the files

<FS>/allegro/etc/email_lists/opt_out
<FS>/allegro/etc/email_lists/opt_in

It is a bit onerous, but you ought to propagate any edits to the files of the same name on all the file systems (FS). Work was begun on a script with GUI

$ALLEGRO/bin/allegro_email_notification.py

but it has not been finished.

High Performance Computing

Allegro’s computers are high performance machines, which are dedicated to reduce ALMA data. However, all users can have access to these machines, although it should be noted that Allegro work has always preference.

As a rule of thumb, other users should be informed if more than 25% of the available cores are needed (i.e., more than 8 cores on chaxa or helada, and 3 cores on tebinquiche). Computational time can be reserved by

$ reserve_cpu_time.sh

This opens up a text file with some explanations of how to “reserve” computing time.

If some high-performance computing time is reserved in the current timeslot, this is displayed through

$ display_cpu_reservations.py

=== High-performance computing on chaxa ===
username  from        to           purpose
-------------------------------------------------------------
alma      26-08-2014  26-08-2014   modelling