diff --git a/RELEASE_NOTES b/RELEASE_NOTES index f2d46a0a65be25ddfcd8fd5f1e2d7506f2e56167..48fa580885805746fbba12cb61f8f2f9e9d1a635 100644 --- a/RELEASE_NOTES +++ b/RELEASE_NOTES @@ -166,8 +166,11 @@ ACCOUNTING CHANGES * A new command, sacctmgr, is available for managing user accounts in SlurmDBD has been added. This information is required for use of SlurmDBD - to manage job accounting data. Additional tools to generate accounting - reports are currently under development and will be released soon. + to manage job accounting data. Information is maintained based upon + an "association", which has four components: cluster name, Slurm partition, + user name and bank account. See the sacctmgr man page or accounting web + page for more information. Additional tools to generate accounting reports + are currently under development and will be released soon. * Job completion records can now be written to a MySQL or PostGreSQL database in addition to a test file as controlled using the JobCompType diff --git a/doc/html/accounting.shtml b/doc/html/accounting.shtml index 52a04b720fb30f0d47e2f4a8c31f8c8f86d33673..0c15b1d433784d0a6b5882f7c52c739d3a01fb9e 100644 --- a/doc/html/accounting.shtml +++ b/doc/html/accounting.shtml @@ -3,9 +3,9 @@ <h1>Accounting</h1> <p>SLURM collects accounting information for every job and job step executed. -This information can be viewed using the <b>sacct</b> command. Information is available about both currently executing jobs and -jobs which have already terminated. +jobs which have already terminated and can be viewed using the +<b>sacct</b> command. Resource usage is reported for each task and this can be useful to detect load imbalance between the tasks. SLURM version 1.2 and earlier supported the storage of accounting @@ -13,7 +13,7 @@ records to a text file. Beginning in SLURM version 1.3 accounting records can be written to a database. </p> -<p>There are three distinct plugins associated with resource accounting. +<p>There are three distinct plugin types associated with resource accounting. The configuration parameters associated with these plugins include: <ul> <li><b>JobCompType</b> controls how job completion information is @@ -24,9 +24,9 @@ information is required, this plugin should satisfy your needs with minimal overhead. You can store this information in a text file, <a href="http://www.mysql.com/">MySQL</a> or <a href="http://www.postgresql.org/">PostgreSQL</a> -database and use either +database optionally using either <a href="http://www.clusterresources.com/pages/products/gold-allocation-manager.php">Gold</a> -or SlurmDBD for added security.</li> +or SlurmDBD for added database security.</li> <li><b>JobAcctGatherType</b> is operating system dependent and controls what mechanisms are used to collect accounting information. Supported values are <i>jobacct_gather/aix</i>, <i>jobacct_gather/linux</i> @@ -35,7 +35,7 @@ and <i>jobacct_gather/none</i> (no information collected).</li> step information is recorded. You can store this information in a text file, <a href="http://www.mysql.com/">MySQL</a> or <a href="http://www.postgresql.org/">PostgreSQL</a> -database and use either +database optionally using either <a href="http://www.clusterresources.com/pages/products/gold-allocation-manager.php">Gold</a> or SlurmDBD for added security.</li> </ul> @@ -64,26 +64,29 @@ daemon can provide better security. Gold and SlurmDBD are two such services. Our initial implementation relied upon Gold, but we found its performance to be inadequate for our needs and developed SlurmDBD. -SlurmDBD (SLURM DataBase Daemon) is written in C, multi-threaded, -secure, and much faster than Gold. +SlurmDBD (SLURM Database Daemon) is written in C, multi-threaded, +secure, and considerably faster than Gold. The configuration required to use SlurmDBD will be described below. Direct database or Gold use would be similar.</p> <p>Note that SlurmDBD relies upon existing SLURM plugins for authentication and database use, but the other SLURM commands and daemons are not required on the host where -SlurmDBD is installed.</p> +SlurmDBD is installed. Install the <i>slurmdbd</i> and +<i>slurm-plugins</i> RPMs on the computer when SlurmDBD +is to execute.</p> <h2>Infrastructure</h2> <p>If the SlurmDBD is executed on a different cluster than the one managed by SLURM, possibly to collect data from multiple -clusters in a single location, then a uniform user ID -space must be provided across all computers (the users need not -be permitted to login to every machine, but every user with an -account on one of the SLURM managed clusters must have an account -on the machine where SlurmDBD executes and that user must have -the same user name and user ID).</p> +clusters in a single location, there are some constraints on +the user space. +The user ID associated with <i>SlurmUser</i> must be uniform +across all clusters. +Accounting is maintained by user name (not user ID), but a +given user name should refer to the same person across all +of the computers.</p> <p>The best way to insure security of the data is by authenticating communications to the SlurmDBD and we recommend @@ -166,7 +169,7 @@ Set to "jobcomp/slurmdbd".</li> This file should be only on the computer where SlurmDBD executes and should only be readable by the user which executes SlurmDBD (e.g. "slurm"). This file should be protected from unauthorized access since it -contains a database password. +contains a database login name and password. See "man slurmdbd.conf" for a more complete description of the configuration parameters. Some of the more important parameters include:</p> @@ -190,7 +193,7 @@ This value must be specified.</li> The port number that the Slurm Database Daemon (slurmdbd) listens to for work. The default value is SLURMDBD_PORT as established at system build time. If none is explicitly specified, it will be set to 6819. -This value must be equal to the \fBSlurmDbdPort\fR parameter in the +This value must be equal to the <i>SlurmDbdPort</i> parameter in the slurm.conf file.</li> <li><b>LogFile</b>: @@ -200,14 +203,14 @@ The default value is none (performs logging via syslog).</li> <li><b>PluginDir</b>: Identifies the places in which to look for SLURM plugins. -This is a colon\-separated list of directories, like the PATH +This is a colon-separated list of directories, like the PATH environment variable. The default value is "/usr/local/lib/slurm".</li> <li><b>SlurmUser</b>: -The name of the user that the \fBslurmctld\fR daemon executes as. +The name of the user that the <i>slurmctld</i> daemon executes as. This user must exist on the machine executing the Slurm Database Daemon -and have the same user ID as the hosts on which \fBslurmctld\fR execute. +and have the same user ID as the hosts on which <i>slurmctld</i> execute. For security purposes, a user other than "root" is recommended. The default value is "root". </li> @@ -235,6 +238,8 @@ Acceptable values at present include "accounting_storage/pgsql". The value "accounting_storage/gold" indicates that account records will be written to Gold, which maintains its own database. +Use of Gold is not recommended due to reduced performance without +providing any additional security. The value "accounting_storage/mysql" indicates that accounting records should be written to a MySQL database specified by the \fStorageLoc\fR parameter. @@ -252,9 +257,20 @@ with to store the job accounting data.</li> <p>Accounting records are maintained based upon what we refer to as an <i>Association</i>, which consists of four elements: -cluster name, partition name, user name, and account name. </p> +cluster name, partition name, user name, and account name. +Use the <i>sacctmgr</i> command to create and manage these records. +You will want to define the names of clusters being managed +by Slurm, the users with accounts on these computers, plus +the user's default and valid account names. Partition names +will be uploaded from Slurm on the cluster, but can be +explicitly defined if so desired.</p> + +<h2>Node State Information</h2> -<p><b>TBD</b></p> +<p>Node state information is also recorded in the database. +Whenever a node goes DOWN or becomes DRAINED that event is +logged along with the node's <i>Reason</i> field. +This can be used to generate various reports.</p> <h2>Tools</h2> @@ -267,6 +283,11 @@ Sacctmgr is used to manage associations in the database: add or remove clusters, add or remove users, etc. See the man pages for each command for more information.</p> -<p style="text-align:center;">Last modified 12 March 2008</p> +<p>Web interfaces with graphical output is currently under +development and should be available in the summer of 2008. +A tool to report node state information is also under development.</p> + + +<p style="text-align:center;">Last modified 19 March 2008</p> <!--#include virtual="footer.txt"-->