QMON(1) | Grid Engine User Commands | QMON(1) |
qmon - X-Windows OSF/Motif graphical user interface for Grid Engine
qmon [options]
Qmon allows administrators and users to manipulate the Grid Engine system from an X-Window environment. Qmon provides various dialogues linked together in multiple ways. For each task the user wishes to accomplish via qmon a corresponding dialogue is provided. There are multiple ways to address the proper dialogue for a certain task:
The supported options are the standard X Toolkit options as described in X(1) section Options. Furthermore, qmon supports:
The Job Control dialogue provides a folder of tabulated lists of the still pending jobs, already running jobs and recently finished jobs. The dialogue allows for detailed information on the jobs as well as for the deletion and suspension of jobs being selected. In addition the job control dialogue offers links to the Submit dialogue in order to submit new jobs or to change attributes of pending jobs (Qalter button). The fields displayed in the tabular display and the jobs displayed can be customized by pressing the Customize button. This customization can be saved to the ~/.qmon_preferences file and is used on following startup for the initial configuration of the Job Control dialogue.
The Queue Control dialogue with its sub-dialogue hierarchy enables the user to control the status of the Grid Engine queues actually configured in the system and allows the administrator to add new queues or to modify or delete existing ones. The Queue Control dialogue shows a tabbed view of Cluster Queues or their corresponding Queue Instances in a tabular way.
By pushing the Add, Clone or Modify button, a sub-dialogue for configuring Grid Engine queues is opened. A queue needs to be selected to use the clone and modify operation. The configuration sub-dialogue allows for definition of the queue and its attributes. The configuration sub-dialogue contains in the upper section the queue name and a list of hosts or hostgroups for which this queue contains the configuration. The lower section contains a list of hosts or hostgroups at the left, where the default attributes are shown when the "@/" entry is highlighted. These default attributes can be overruled on a hostgroup or host basis by enabling attributes and changing their values after this hostgroup or host entry has been added in the "Attributes for Host/Hostgroup" listbox. The queue configuration parameters (see queue_conf(5)) are subdivided into different categories (General Configuration, Execution Methods, Checkpointing, Load/Suspend Thresholds, Limits, Complexes, User Access, Project Access, Subordinate Queues, Owners) which are selectable by the tab widget area presented in the lower region of the queue configuration sub-dialogue. The administrator may select previously-configured settings from already existing queues (Clone button). By pushing the Ok button, the definitions are registered with sge_qmaster(8). The Queue Control dialogue can be customized in a similar way as the Job Control dialogue. The settings applied here are also saved in ~/.qmon_preferences.
The Job Submission dialogue serves for submitting batch and interactive jobs and is also invoked when changing attributes of pending jobs from the Job Control dialogue explained above (Qalter button). To toggle between batch and interactive jobs please use the Batch/Interactive button at the top of the button column on the right side of the Job Submission screen.
The dialogue consists of a folder containing two job preparation dialogue pages. The most frequently used parameters in the course of a job submission are offered on the General page. A job script has to be defined; all other fields are optional. If the job demands specification of advanced requirements, the Advanced tab can be used to switch to an enhanced parameter display.
If resource requirements are mandatory for the job, the Request Resources icon button has to be used to pop up the Requested Resources sub-dialogue. This sub-dialogue allows for selection of the required resources of the job and for definition of the quantities in which this resources are to be provided. The Available Resources are constituted from those complex attributes declared requestable (see complex(5) for details). Resource requirements can be made Hard, i.e. they must be met before a job can be started in a queue, or Soft, i.e. they are granted on an as-available basis.
Closing the Requested Resources sub-dialogue with the done button books the specified requirement for the job. Pushing the Submit button on the top level Submit dialogue submits the job.
The Complex Config allows the administrator to add new complex attributes or to modify or delete existing ones (see complex(5)). The complex configuration dialogue provides a tabulated list of the complex attribute entries and an input region for the declaration of new or modified entries. The Add button updates the tabulated list with a new attribute. The Modify button changes a highlighted existing entry. The Delete button removes the highlighted attributes. The Load and Save buttons allow saving or reloading the whole attribute list to or from file. The Ok button registers the additional or modified complex attributes with sge_qmaster(8).
Four types of host related lists can be maintained via the Host Config dialogue:
Administration Hosts Submit Hosts Host Groups Execution Hosts
The host list to be manipulated is selected via clicking at one of the tabs named correspondingly. The first two host lists only provide for adding or deleting entries, thereby allowing administrative or submit permission for the hosts on the lists, or denying it otherwise respectively. The host group list allows the manipulation of host groups. Host groups can reference either other host groups or hosts. The execution host list entries in addition provide the ability to define scaling factors for the load sensors, consumable complex attributes and access attributes (access, xaccess and projects, xprojects) as described in complex(5). CPU, memory and I/O usage reported for running jobs can be scaled in addition and the relative performance of a host can be defined with the Resource Capability Factor (see host_conf(5)).
This dialogue maintains the cluster global configuration as well as host specific derivatives (see sge_conf(5)). When opened, the dialogue displays a selection list for all hosts which have a configuration assigned. The special name "global" refers to the cluster global configuration. By pushing the Add/Modify button a sub-dialogue is opened, which allows for modification of the cluster configuration. For host specific configurations the 'global' host specific configuration fields are set insensitive and only the allowed parameters can be manipulated.
The Scheduler Configuration dialogue provides the means to change the behavior of the Grid Engine scheduler thread part of sge_qmaster(8) process. The dialogue contains a representation for all scheduler configuration parameters as described in sched_conf(5). It is subdivided in the two sections General Parameters and Load Adjustments, which can be selected via the corresponding tabs. The Ok button registers any changes with sge_qmaster(8).
The Calendar Config allows the administrator to add new calendars or to modify or delete existing ones (see calendar_conf(5)). The dialogue offers a selection list for the existing calendars and displays the configuration of the one being selected. By pushing the Delete button, the selected calendar is deleted from the configuration. Pushing the Add/Modify button will open a calendar configuration dialogue, which allows to create new calendars or which provides the means to change the existing ones. The Ok button registers the additional or modified calendar with sge_qmaster(8).
User permissions are controlled via the User Config dialogue. The tab widget in the left section of the dialogue allows for selecting between
Configuration of Manager accounts. Configuration of Operator accounts. Definition of Usersets. Definition of User accounts.
Those user accounts added to the list of manager or operator accounts are given permission to act as managers or operators respectively when accessing Grid Engine under their own account.
The userset lists are used together with the user_lists and xuser_lists host, queue, project and cluster configuration parameters (see queue_conf(5), project(5) and sge_conf(5)) to control access of users to hosts, queues, projects and the entire cluster. A userset is just a collection of user names and UNIX group names. Group names are identified by prefixing them with a "@" sign. The already-defined usersets are displayed in a selection list. These lists can be modified, and new lists can be created, using the Userset definition dialogue.
Usersets can be used as Access List and/or as Department required for the so called Functional Policy and Override Policy (see Ticket Config below).
User names can be added to the system as entries to the Grid Engine user database (see user(5)); here a default project can be attached to the user too. This can be done with the User sub-dialogue.
The Tickets button in the button list on the right side of the dialogue opens the Ticket Config dialogue (see below).
Parallel environment (PE) interfaces can be configured with this dialogue. PE interfaces are necessary to describe how parallel programming environments like PVM (Parallel Virtual Machine), MPI (Message Passing Interface), or shared memory parallel systems, are to be instantiated and to impose access restrictions onto the PEs. When the dialogue is opened a list of the configured PEs is displayed, together with the current configuration (see sge_pe(5)) of the selected PE interface. To add new PE interfaces, or to modify existing ones, an Add and a Modify button are available which open a PE interface configuration sub-dialogue. After applying the changes and quitting this sub-dialogue with the OK button, the new or modified PE interface is registered with sge_qmaster(8).
Checkpointing environment interfaces can be configured with this dialogue. Checkpointing environments are necessary to describe the attributes which the different checkpointing methods and their derivatives on various operating system platforms supported by Grid Engine have. When the dialogue is opened a list of the already configured checkpointing environments is displayed together with the current configuration (see checkpoint(5)) of the selected checkpointing environment. To add new checkpointing environments, or to modify existing ones, an Add and a Modify button is available which opens a checkpointing environment configuration sub-dialogue. After applying the changes and quitting this sub-dialogue with the OK button, the new or modified checkpointing environment is registered with sge_qmaster(8).
This dialogue offers an overview and editing screen for allocating tickets to the share-based, functional and override scheduling policies.
The Deadline Job button opens the User Conf dialogue box. Please change to the Userset sub-dialogue and select the userset named "deadlineusers". Only users of this userset may submit deadline jobs.
The Share Tree Policy button opens the dialogue for creating and editing the Grid Engine share tree (see share_tree(5) and sched_conf(5) for a description of the configuration parameters).
The Functional Policy button opens the dialogue for creating and editing the allocation of the functional shares (see sched_conf(5), access_list(5), project(5), queue_conf(5) and user(5) for a description of the different types of functional shares and the configurable weighting parameters).
The Override Policy button opens the dialogue for creating and editing the allocation of override tickets (see access_list(5), project(5), queue_conf(5) and user(5) for a description of the different types of override tickets).
This button opens a dialog for creating projects.
The dialogue offers a selection list for the existing projects and displays
the configuration of the one being selected. By pushing the Delete
button, the selected project is deleted from the configuration. Pushing the
Add/Modify button will open a project configuration dialogue,
which allows creating new projects changing existing ones. Project
configuration in essence means giving or denying access to a project for
usersets (see User Conf above as well as project(5)). The Ok
button registers the additional or modified project with
sge_qmaster(8).
The Object Browser dialogue's purpose is manifold: First of all, Grid Engine and qmon Messages, such as notification of error or success concerning a previously taken action, can be displayed in the dialogue's output window. Also the standard output and the standard error output of qmon can be diverted to the Object Browser output window.
Additionally the Object Browser can be used to display continuous information about Grid Engine objects as the mouse pointer moves over their representation as icons or table entries in other qmon dialogues. Currently, only the display of the configuration of two Grid Engine objects in two separate dialogues is supported:
The Exit icon button is not linked with a dialogue. Its sole purpose is to close all active qmon dialogues and to exit the application.
The available resources, their meaning and the syntax to be followed in order to modify them are described in the default qmon resource file (see the section Files below for the location of the resource file).
The name of the default cell, i.e. default.
If the line to be entered in an editing window is longer than the width of the window, then the text just runs off the end of the window.
<sge_root>/qmon/Qmon Qmon sample resources file /usr/lib/X11/defaults/Qmon Qmon system resources file $HOME/Qmon Qmon user resources file $HOME/.qmon_preferences Qmon job/queue customization file
sge_intro(1), sge_conf(5), access_list(5), sge_pe(5), calendar_conf(5), complex(5), project(5), queue_conf(5), sched_conf(5), user(5), sge_qmaster(8).
See sge_intro(1) and the information provided in <sge_root>/3rd_party/qmon for a statement of further rights and permissions and for credits to be given to public domain and freeware widget developers.
2011-05-18 | SGE 8.1.3pre |