Differences between revisions 51 and 52
Revision 51 as of 2019-06-04 11:11:25
Size: 16759
Revision 52 as of 2019-06-04 11:17:21
Size: 16863
Deletions are marked like this. Additions are marked like this.
Line 55: Line 55:
Accounting records are maintained based upon what we refer to as an Association_, which consists of four elements: Accounting records are maintained based upon what we refer to as an Association_, which is a **fundamental concept** consisting of four elements:
Line 63: Line 63:

Display all Association_ records by::

  sacctmgr show associations

Slurm accounting

Jump to our Slurm top-level page.

The following configuration is relevant for the Head/Master node only.

Accounting setup in Slurm

See the accounting page and the Slurm_tutorials with Slurm Database Usage.

Before setting up accounting, you need to set up the Slurm_database.

Database Configuration

See the section Database Configuration in the accounting page. The command to manage accounts is:

  • sacctmgr - View and modify Slurm account information


Accounting records are maintained based upon what we refer to as an Association, which is a fundamental concept consisting of four elements:

  • cluster,
  • account,
  • user names,
  • an optional partition name.

Use the sacctmgr command to create and manage the Association records.

Display all Association records by:

sacctmgr show associations

NOTE: There is an order to set up accounting associations. You must define clusters before you add accounts, and you must add accounts before you can add users.

Create a cluster

Add the cluster name to slurm.conf:


See the Database Configuration section of the accounting page:

sacctmgr add cluster niflheim
sacctmgr create cluster niflheim  # The add and create commands are identical

See also https://bugs.schedmd.com/show_bug.cgi?id=1050

Create accounts and users

See the Database Configuration section of the accounting page.

Create accounts

Definition of accounts:

  • Not formally defined in the Slurm documentation, but see the accounting page for examples.
  • An account is similar to a UNIX group.
  • An account may contain multiple users, or just a single user.
  • Accounts may be organized as a hierarchical tree.
  • A user may belong to multiple accounts, but must have a DefaultAccount (see sacctmgr).

Create a hierarchical organization list using sacctmgr, for example, with departments and external users:

sacctmgr add account dtu Description="DTU departments" Organization=dtu

sacctmgr add account fysik Description="Physics department" Organization=fysik parent=dtu
sacctmgr add account deptx Description="X department" Organization=deptx parent=dtu

sacctmgr add account external Description="External groups" Organization=external
sacctmgr add account companyx Description="Company X" Organization=companyx parent=external

You may also create subgroups within the departments:

sacctmgr add account camd Description="CAMD section" Organization=camd parent=fysik

If you wish to assign different resources within the departmental subgroups, you could use the UNIX GID group name to differentiate between faculty, staff and students, for example. Use the GID names from the /etc/group file to create new accounts within the same Organization name, for example, for the CAMD section students with GID group name camdstud:

sacctmgr add account camdstud Description="CAMD students" Organization=camd parent=camd

Display the accounts created:

sacctmgr show account
sacctmgr show account -s   # Show also associations in the accounts

When either adding or modifying an account, the following sacctmgr options are available:

  • Cluster= Only add this account to these clusters. The account is added to all defined clusters by default.
  • Description= Description of the account. (Default is account name)
  • Name= Name of account. Note the name must be unique and can not represent different bank accounts at different points in the account hierarchy
  • Organization= Organization of the account. (Default is parent account unless parent account is root then organization is set to the account name.)
  • Parent= Make this account a child of this other account (already added).

Create users

Create a Slurm user named xxx with a specific default account (required) yyy:

sacctmgr create user name=xxx DefaultAccount=yyy

If desired users may also be added to additional accounts (see accounting), for example:

sacctmgr add user xxx Account=zzzz

The fairshare and other settings for the non-default account may be configured:

sacctmgr modify user where name=xxx account=zzzz set fairshare=0

A non-default account name may be specified in the user's batch jobs, for example with sbatch:

sbatch -A <account> or --account=<account>

List users by:

sacctmgr show user
sacctmgr show user -s
sacctmgr show account -s xxx

When either adding or modifying a user, the following sacctmgr options are available:

  • Account= Account(s) to add user to (see also DefaultAccount).
  • AdminLevel= This field is used to allow a user to add accounting privileges to this user. Valid options are:
    • None
    • Operator: can add, modify, and remove any database object (user, account, etc), and add other operators. On a SlurmDBD served slurmctld these users can:
      • View information that is blocked to regular uses by a PrivateData flag (see slurm.conf).
      • Create/Alter/Delete Reservations
    • Admin: These users have the same level of privileges as an operator in the database. They can also alter anything on a served slurmctld as if they were the slurm user or root.
  • Cluster= Only add to accounts on these clusters (default is all clusters)
  • DefaultAccount= Default account for the user, used when no account is specified when a job is submitted. (Required on creation)
  • DefaultWCKey= Default WCkey for the user, used when no WCkey is specified when a job is submitted. (Only used when tracking WCkey.)
  • Name= User name
  • Partition= Name of Slurm partition this association applies to.
Modifying user UIDs or group GIDs

If a user's UID is modified in /etc/passwd, or the user's GID is modified in /etc/passwd, the Slurm design requires you to restart slurmctld, see bug_3575.

User fairshare

By default users are created with a raw fairshare value of 1, see the sshare command:

sshare -l -a

One may alternatively create or modify users with a non-default fairshare value, see Resource_Limits or sacctmgr:

  • Fairshare= Integer value used for determining priority. Essentially this is the amount of claim this association and it's children have to the above system. Can also be the string parent, when used on a user this means that the parent association is used for fairshare. If Fairshare=parent is set on an account, that account's children will be effectively re-parented for fairshare calculations to the first parent of their parent that is not Fairshare=parent. Limits remain the same, only it's fairshare value is affected.

NOTE: The Fairshare=parent definition is inconsistent, since the Multifactor_Priority_Plugin says something different:

  • If all users in an account is configured with FairShare=parent the result is that all the jobs drawing from that account will get the same fairshare priority, based on the accounts total usage. No additional fairness is added based on users individual usage.

For example:

sacctmgr create user name=xxx DefaultAccount=yyy fairshare=2

Modifying and removing entities

For example, to permit user xxx to execute jobs on all clusters with a default account of fysik execute:

sacctmgr add user xxx DefaultAccount=fysik

You can modify the database items using SQL-like where and set, for example:

sacctmgr modify account where cluster=niflheim name=fysik set Description="DTU Physics"

The following has been copied from the accounting page:

When modifying entities, you can specify many different options in SQL-like fashion, using key words like where and set. A typical execute line has the following form:

sacctmgr modify <entity> set <options> where <options>

For example:

sacctmgr modify user set default=none where default=test

will change all users with a default account of "test" to account "none". Once an entity has been added, modified or removed, the change is sent to the appropriate Slurm daemons and will be available for use instantly.

Removing entities using an execute line similar to the modify example above, but without the set options. For example, remove all users with a default account "test" using the following execute line:

sacctmgr remove user where default=test

To remove a user from an account:

sacctmgr remove user brian where account=physics

Note: In most cases, removed entities are preserved, but flagged as deleted. If an entity has existed for less than 1 day, the entity will be removed completely. This is meant to clean up after typographic errors.

Resource Limits

To enable any limit enforcement you must at least have:


in your slurm.conf, otherwise, even if you have limits set, they will not be enforced. Other options for AccountingStorageEnforce and the explanation for each are found on the Resource_Limits document.

Now you can impose user limits, for example:

sacctmgr modify user xxx set GrpTRES=cpu=1000 GrpTRESRunMin=cpu=2000000

Notice: The keyword cpu must be in lower-case, see bug_4226 (resolved in Slurm 17.02.8).

Trackable RESources (TRES)

A TRES is a resource that can be tracked for usage or used to enforce limits against.

Quality of Service (QOS)

A Quality of Service (QOS) configuration should have at least the default QOS value of normal. Additional QOS'es may be defined, for example a high QOS with some priority and limits (see sacctmgr):

sacctmgr add qos high priority=10 MaxTRESPerUser=cpu=256

The Quality of Service (QOS) Factor is defined in the Multifactor_Priority_Plugin page as:

Each QOS can be assigned an integer priority. The larger the number, the greater the job priority will be for jobs that request this QOS. This priority value is then normalized to the highest priority of all the QOS's to become the QOS factor.

View defined QOS'es with various degrees of detail by:

sacctmgr show qos
sacctmgr show qos format=name
sacctmgr --noheader show qos format=name

A user may be allowed to use a certain QOS like in these examples:

sacctmgr -i modify user where name=XXXX set QOS=normal,high
sacctmgr -i modify user where name=XXXX set QOS+=high

The user DefaultQOS (see sacctmgr) may be set by:

sacctmgr -i modify user where name=XXXX set DefaultQOS=normal

Users must submit jobs to non-default QOS with sbatch like:

sbatch --qos=high ...

Workload Characterization Key (WCKey)

A WCKey is an orthogonal way to do accounting against possibly unrelated accounts. This can be useful where users from different accounts are all working on the same project.

Enforce accounting

The account and user associations created above only take effect after you enable:


in your slurm.conf. Options for AccountingStorageEnforce and the explanation for each are found in slurm.conf and the Resource_Limits document.

When AccountingStorageEnforce is changed, a restart of the slurmctld daemon is required (not just a scontrol reconfig).

Accounting information

Inquire accounting information using these commands:

  • sacct - Displays accounting data for all jobs and job steps in the Slurm job accounting log or Slurm database.
  • sstat - Display various status information of a running job/step.
  • sreport - Generate reports from the slurm accounting data.
  • scontrol show assoc_mgr - displays the current contents of the slurmctld 's internal cache for users, associations and/or qos.

Useful commands

  • Show user fairshare etc. information:

    sacctmgr show associations format=account,user,fairshare,GrpTRES,GrpTRESRunMin

Slurm job efficiency report (seff)

The /usr/bin/seff command takes a jobid and reports on the efficiency of that job's cpu and memory utilization (requires Slurm 15.08 or later). The slurm-contribs RPM (Slurm 17.02 and later, previously slurm-seff) also comes with an /usr/bin/smail utility that allows for Slurm end-of-job emails to include a seff report, see bug_1611. This allows users to become aware if they are wasting resources.

Note: You may like to copy the updated smail from https://github.com/OleHolmNielsen/Slurm_tools/tree/master/smail to add the cluster name to mail headers. This is included in Slurm 17.11, but make sure to get the bugfix in 17.11.6.

The smail utility is invoked automatically to process end-of-job notifications if you add the following to slurm.conf:


User job scripts may also use this line as the last line:


Accounting reports

Use sreport to generate reports from the slurm accounting data, for example:

sreport cluster UserUtilizationByAccount
sreport cluster AccountUtilizationByUser

The accounting timings will by default be displayed in units of TRES Minutes.

Selection of date ranges:

sreport ... Start=02/01 End=02/25
sreport ... Start=`date -d "last month" +%D` End=`date -d "this month" +%D`

Change the date/time format in report header for readability (formats in "man strftime"):

env SLURM_TIME_FORMAT="%d-%b-%Y_%R" sreport ...

Show accounting indented as a tree:

sreport cluster AccountUtilizationByUser tree

Show top user accounting:

sreport user top start=0101 end=0201 TopCount=50 -t hourper --tres=cpu,gpu

Specify the accounting time format (default is Minutes) from sreport:

sreport -t hourper ...

Report specified TRES accounting (default is cpu):

sreport --tres cpu,gpu ...

Print parseable output from sreport:

sreport -p ...

for further processing with scripts.

Cluster utilization report:

sreport -t hourper cluster Utilization

Niflheim: Slurm_accounting (last edited 2019-09-25 08:10:26 by OleHolmNielsen)