MMM Points of Contact
User management tools§
This page contains tools and information for the nominated Points of Contact.
Other system-specific information is at Michael or Young.
These commands can all be run as
michael-command or
young-command:
they run the same thing and the different names are for convenience.
Displaying user information§
young-show, michael-show or young-show is a tool that enables you to find a lot
of information about users. Access to the database is given to points of
contact individually, contact rc-support@ucl.ac.uk if you try to use
this and get an access denied.
At the top level, --user shows all information for one user, in
multiple tables. --contacts shows all points of contact - useful for
getting the IDs, and --institutes is the same. --allusers will show
everyone's basic info. --getmmm will show the most recently used mmm
username.
young-show -h
usage: young-show [-h] [--user username] [--contacts] [--institutes]
[--allusers] [--getmmm]
{recentusers,getusers,whois} ...
Show data from the Young database. Use [positional argument -h] for more
help.
positional arguments:
{recentusers,getusers,whois}
recentusers Show the n newest users (5 by default)
getusers Show all users with this project, institute, contact
whois Search for users matching the given requirements
optional arguments:
-h, --help show this help message and exit
--user username Show all current info for this user
--contacts Show all allowed values for contact
--institutes Show all allowed values for institute
--allusers Show all current users
--getmmm Show the highest mmm username used
Show recent users§
young-show recentusers shows you the
most recently-added N users, default 5.
young-show recentusers -h
usage: young-show recentusers [-h] [-n N]
optional arguments:
-h, --help show this help message and exit
-n N
Show users with a given project, institute, contact§
young-show getusers will search for exact
matches to the given project, institute, contact combination.
young-show getusers -h
usage: young-show getusers [-h] [-p PROJECT] [-i INST_ID] [-c POC_ID]
optional arguments:
-h, --help show this help message and exit
-p PROJECT, --project PROJECT
Project name
-i INST_ID, --institute INST_ID
Institute ID
-c POC_ID, --contact POC_ID
Point of Contact ID
Search for users based on partial information§
young-show whois can be used to search for
partial matches to username, name, email fragments, including all of
those in combination.
young-show whois -h
usage: young-show whois [-h] [-u USERNAME] [-e EMAIL] [-n GIVEN_NAME]
[-s SURNAME]
optional arguments:
-h, --help show this help message and exit
-u USERNAME, --user USERNAME
UCL username of user contains
-e EMAIL, --email EMAIL
Email address of user contains
-n GIVEN_NAME, --name GIVEN_NAME
Given name of user contains
-s SURNAME, --surname SURNAME
Surname of user contains
Adding user information and new projects§
young-add will add information to the database.
Access to the database is given to points of contact individually,
contact rc-support@ucl.ac.uk if you try to use this and get an access
denied.
Please note that all options have a --debug flag that will allow you
to see the query generated without committing the changes to the
database - double-check that the information you are adding is
correct.
young-add -h
usage: young-add [-h] {user,project,projectuser,poc,institute} ...
Add data to the Young database. Use [positional argument -h] for more help.
positional arguments:
{user,project,projectuser,poc,institute}
csv Add all users from the provided CSV file
user Adding a new user with their initial project
project Adding a new project
projectuser Adding a new user-project-contact relationship
poc Adding a new Point of Contact
institute Adding a new institute/consortium
optional arguments:
-h, --help show this help message and exit
Add a new user§
young-add user allows you to add a new user,
with their initial project and point of contact. As of 27 June 2022 this now goes
ahead and creates their account automatically within 10 minutes - first prompting you
that the information you have entered is correct. You do not need to email us separately
about creating accounts unless something has gone wrong. The user's initial project
must already exist (create with young-add project first).
The user will be allocated the next free mmmxxxx username - you should
only specify username yourself if they are an existing UCL user, or on
Young if they previously had a Michael account you should give
them the same username. If they already have an account on this cluster with
a different institution, just add them as a projectuser instead using their
existing username.
You can get your poc_id by looking at young-show --contacts.
young-add user -h
usage: young-add user [-h] -u USERNAME -n GIVEN_NAME [-s SURNAME] -e
EMAIL_ADDRESS -k "SSH_KEY" -p PROJECT_ID -c POC_ID
[--debug]
optional arguments:
-h, --help show this help message and exit
-u USERNAME, --user USERNAME
UCL username of user
-n GIVEN_NAME, --name GIVEN_NAME
Given name of user
-s SURNAME, --surname SURNAME
Surname of user (optional)
-e EMAIL_ADDRESS, --email EMAIL_ADDRESS
Institutional email address of user
-k "SSH_KEY", --key "SSH_KEY"
User's public ssh key (quotes necessary)
-p PROJECT_ID, --project PROJECT_ID
Initial project the user belongs to
-c POC_ID, --contact POC_ID
Short ID of the user's Point of Contact
--noconfirm Don't ask for confirmation on user account creation
--verbose Show SQL queries that are being submitted
--debug Show SQL query submitted without committing the change
SSH key formats§
It will verify the provided ssh key by default. Note that it has to be
in the form ssh-xxx keystartshere. If someone has sent in a key which
has line breaks and header items, make it into this format by adding the
key type to the start and removing the line breaks from the key body.
This key:
---- BEGIN SSH2 PUBLIC KEY ----
Comment: "comment goes here"
AAAAB3NzaC1yc2EAAAABJQAAAQEAlLhFLr/4LGC3cM1xgRZVxfQ7JgoSvnVXly0K
7MNufZbUSUkKtVnBXAOIjtOYe7EPndyT/SAq1s9RGZ63qsaVc/05diLrgL0E0gW+
9VptTmiUh7OSsXkoKQn1RiACfH7sbKi6H373bmB5/TyXNZ5C5KVmdXxO+laT8IdW
7JdD/gwrBra9M9vAMfcxNYVCBcPQRhJ7vOeDZ+e30qapH4R/mfEyKorYxrvQerJW
OeLKjOH4rSnAAOLcEqPmJhkLL8k6nQAAK3P/E1PeOaB2xD7NNPqfIsjhAJLZ+2wV
3eUZATx9vnmVF0YafOjvzcoK2GqUrhNAvi7k0f+ihh8twkfthj==
---- END SSH2 PUBLIC KEY ----
should be converted into
ssh-rsa AAAAB3NzaC1yc2EAAAABJQAAAQEAlLhFLr/4LGC3cM1xgRZVxfQ7JgoSvnVXly0K7MNufZbUSUkKtVnBXAOIjtOYe7EPndyT/SAq1s9RGZ63qsaVc/05diLrgL0E0gW+9VptTmiUh7OSsXkoKQn1RiACfH7sbKi6H373bmB5/TyXNZ5C5KVmdXxO+laT8IdW7JdD/gwrBra9M9vAMfcxNYVCBcPQRhJ7vOeDZ+e30qapH4R/mfEyKorYxrvQerJWOeLKjOH4rSnAAOLcEqPmJhkLL8k6nQAAK3P/E1PeOaB2xD7NNPqfIsjhAJLZ+2wV3eUZATx9vnmVF0YafOjvzcoK2GqUrhNAvi7k0f+ihh8twkfthj==
Other types of keys (ed25519 etc) will say what they are in the first line,
and you should change the ssh-rsa appropriately. The guide linked at
Creating an ssh key in
Windows
also shows where users can get the second format out of PuTTY.
Add new users in bulk from a CSV file§
young-add csv allows you to add users in bulk using a CSV file of specific format
and headers. As of 27 June 2022 the accounts will be all created and activated
automatically within 10 minutes.
The CSV is comma-separated with a header line of
email,given_name,surname,username,project_ID,ssh_key
You can leave username empty for it to allocate them a new username, but if they have an existing mmm username you should fill it in. It may be useful to show users with a given institute on Young if you are migrating users from one service to another.
You can download a CSV template here. Replace the example data.
young-add csv will try to automatically get your Point of Contact ID based on
your username. If it can't, or if you have more than one, it will give you a
list to choose from. (All users in one CSV upload will be added using the
same Point of Contact ID).
It will prompt you for confirmation on each user account creation unless you give
the --noconfirm option.
The project you are adding the user to must already exist.
The SSH key must be formatted as shown in SSH key formats.
If you check your CSV file on the cluster with cat -v and it shows that it is
beginning with M-oM-;M-? and ending with ^M you probably need to run
dos2unix on it first.
Add a new project§
young-add project will create a new project,
associated with an institution. It will not show in Gold until it also
has a user in it.
A project ID should begin with your institute ID, followed by an underscore and a project name.
young-add project -h
usage: young-add project [-h] -p PROJECT_ID -i INST_ID [--debug]
optional arguments:
-h, --help show this help message and exit
-p PROJECT_ID, --project PROJECT_ID
A new unique project ID
-i INST_ID, --institute INST_ID
Institute ID this project belongs to
--debug Show SQL query submitted without committing the change
Add a new project/user pairing§
young-add projectuser will add an
existing user to an existing project. Creating a new user for an
existing project also creates this relationship. After a new
project-user relationship is added, a cron job will pick that up within
15 minutes and create that project for that user in Gold, with no
allocation.
young-add projectuser -h
usage: young-add projectuser [-h] -u USERNAME -p PROJECT_ID -c POC_ID
[--debug]
optional arguments:
-h, --help show this help message and exit
-u USERNAME, --user USERNAME
An existing UCL username
-p PROJECT_ID, --project PROJECT_ID
An existing project ID
-c POC_ID, --contact POC_ID
An existing Point of Contact ID
--debug Show SQL query submitted without committing the change
Deactivating information§
This tool is only partly functional at present. It allows you to deactivate (not delete) some entities that may no longer exist or may have been created in error.
Deactivate a projectuser§
Use this when the user should no longer be a member of the given project. It
does not deactivate the user account, just their membership in this project.
You can confirm the change by looking at young-show --user - it will say
'deactivated' rather than 'active' next to their listing for this project.
young-deactivate projectuser -h
usage: young_deactivate.py projectuser [-h] -u USERNAME -p PROJECT [--debug]
optional arguments:
-h, --help show this help message and exit
-u USERNAME, --user USERNAME
An existing username
-p PROJECT, --project PROJECT
An existing project ID
--debug Show SQL query submitted without committing the change
Gold resource allocation§
We are currently using Gold to manage allocations. The Michael and Young clusters use separate databases for this, so projects on one will not appear on the other.
Reporting from Gold§
There are wrapper scripts for a number of Gold commands (these exist in
the userscripts module, loaded by default).
These are all set to report in cpu-hours with the -h flag, as that is
our main unit. If you wish to change anything about the wrappers, they
live in /shared/ucl/apps/cluster-scripts/ so you can take a copy and
add your preferred options.
They all have a --man option to see the man pages for that command.
Here are some basic useful options and what they do. They can all be given more options for more specific searches.
gusage -p project_name [-s start_time] # Show the Gold usage per user in this project, in the given timeframe if specified.
gbalance # Show the balance for every project, split into total, reserved and available.
glsuser # Shows all the users in Gold.
glsproject # Shows all the projects and which users are in them.
glsres # Show all the current reservatioms, inc user and project. The Name column is the SGE job ID.
gstatement # Produce a reporting statement showing beginning and end balances, credits and debits.
# Less useful commands
glstxn # Show all Gold transactions. Filter or it will take forever to run.
glsalloc # Show all the allocations.
These can be run by any user. The date format is YYYY-MM-DD.
Eg. gstatement -p PROJECT -s 2017-08-01 will show all credits and
debits for the given project since the given date, saying which user and
job ID each charge was associated with.
Transferring Gold§
As the point of contact, you can transfer Gold from your allocation
account into other project accounts. As before, we've put -h in the
wrapper so it is always working in cpu-hours.
gtransfer --fromProject xxx_allocation --toProject xxx_subproject cpu_hours
You can also transfer in the opposite direction, from the subproject back into your allocation account.
Note that you are able to transfer your allocation into another institute's projects, but you cannot transfer it back again - only the other institute's point of contact (or rc-support) can give it back, so be careful which project you specify.
When two allocations are active§
There is now an overlap period of a week when two allocations can be
active. By default, gtransfer will transfer from active allocations in
the order of earliest expiring first. To transfer from the new
allocation only, you need to specify the allocation id.
gtransfer -i allocation_ID --fromProject xxx_allocation --toProject xxx_subproject cpu_hours
glsalloc -p xxx_allocation shows you all allocations that ever existed for your institute,
and the first column is the id.
Id Account Projects StartTime EndTime Amount Deposited Description
--- ------- --------------------- ---------- ---------- ---------- ---------- --------------
87 38 UKCP_allocation 2017-08-07 2017-11-05 212800.00 3712800.00
97 38 UKCP_allocation 2017-10-30 2018-02-04 3712800.00 3712800.00