| Previous | Contents | Index |
The MONITOR PAGE command initiates monitoring of the PAGE class.
MONITOR PAGE
/qualifier[,...]
One or more qualifiers as described in the Command Qualifier Descriptions section.
/ALL
Specifies that a table of all available statistics (current, average, minimum, and maximum) is to be included in the display and summary output. For summary output, this qualifier is the default for all classes; otherwise, it is the default for all classes except CLUSTER, MODES, PROCESSES, STATES, SYSTEM, and VECTOR./AVERAGE
Specifies that a bar graph of average statistics is to be included in the display and summary outputs./CURRENT
Specifies that a bar graph of current statistics is to be included in the display and summary outputs. The /CURRENT qualifier is the default for the CLUSTER, MODES, STATES, SYSTEM, and VECTOR classes./MAXIMUM
Specifies that a bar graph of maximum statistics is to be included in the display and summary outputs./MINIMUM
Specifies that a bar graph of minimum statistics is to be included in the display and summary outputs.
The PAGE class includes the following data items:
Data Item Description Page Fault Rate Rate of page faults for all working sets Page Read Rate Rate of pages read from disk as a result of page faults Page Read I/O Rate Rate of read I/O operations from disk as a result of page faults Page Write Rate Rate at which pages were written to the page file Page Write I/O Rate Rate of write I/O operations to the page file Free List Fault Rate Rate at which pages were read from the free-page list as a result of page faults Modified List Fault Rate Rate of pages read from the modified-page list as a result of page faults Demand Zero Fault Rate Rate at which zero-filled pages were allocated as a result of page faults Global Valid Fault Rate Rate of page faults for pages that are not in the process's working set, but are in physical memory and are indicated as valid pages in the systemwide global page tables Writes In Progress Fault Rate Rate of pages read that were in the process of being written back to disk when faulted System Fault Rate Rate of page faults for pages in system space Free List Size Number of pages on the free-page list Modified List Size Number of pages on the modified-page list
MONITOR> MONITOR PAGE
|
This example shows that the current rate of pages read per read I/O operation is approximately 3 per second (Page Read Rate divided by Page Read I/O Rate). Note that while the page fault rate is currently at the highest point of the monitoring session, the majority of the pages are faulted from memory, not from disk.
The MONITOR PROCESSES command initiates monitoring of the PROCESSES class, which displays information about all processes in the system.In a multifile summary request, the classes CLUSTER and PROCESSES are ignored. If these classes are the only classes specified on the command line, MONITOR does not recognize them and displays a "no classes specified" error message.
Beginning in OpenVMS Version 8.3, four new qualifiers ( /TOPKERNEL, /TOPEXECUTIVE, /TOPSUPERVISOR, and /TOPUSER) allow you to monitor per-process-based modes usage. These qualifiers are useful in helping to identify the top consumers of the various CPU modes. For example, if the MONITOR MODES command reveals that an excessive amount of supervisor mode is being used, the new MONITOR PROCESSES/TOPSUPERVISOR display reveals which process---and therefore, which user---is responsible.
MONITOR PROCESSES
/qualifier[,...]
One or more qualifiers as described in the Command Qualifier Descriptions section.
/TOPBIO
Specifies that a bar graph listing the top buffered I/O users be produced instead of the standard display and summary output. Values are expressed in units of buffered I/Os per second./TOPCPU
Specifies that a bar graph listing the top CPU time users be produced instead of the standard display and summary output. Values are expressed in units of clock ticks (10 milliseconds) per second.Prior to OpenVMS Version 7.3, the MONITOR PROCESSES/TOPCPU display showed only a maximum of 8 processes on one screen. In OpenVMS Version 7.3 and later versions, the choice of which one of three screens is displayed is determined by the number of CPUs on the system. (See the examples in this section.)
/TOPDIO
Specifies that a bar graph listing the top direct I/O users be produced instead of the standard display and summary output. Values are expressed in units of direct I/Os per second./TOPEXECUTIVE
Specifies that a bar graph listing the top executive-mode users be produced instead of the standard display and summary output. Values are expressed in clock ticks (10 ms) per second./TOPFAULT
Specifies that a bar graph listing the top page-faulting processes be produced instead of the standard display and summary output. Values are expressed in number of page faults per second./TOPKERNEL
Specifies that a bar graph listing the top kernel-mode users be produced instead of the standard display and summary output. Values are expressed in clock ticks (10 ms_ per second./TOPRBS (VAX only)
On OpenVMS VAX systems, specifies that a bar graph listing the top balance slot faulting processes be produced instead of the standard display and summary output. Values are expressed in balance slot faults per second./TOPSUPERVISOR
Specifies that a bar graph listing the top supervisor-mode users be produced instead of the standard display and summary output. Values are expressed in clock ticks (10 ms) per second./TOPUSER
Specifies that a bar graph listing the top user-mode users be produced instead of the standard display and summary output. Values are expressed in clock ticks (10 ms_ per second.
As illustrated in the examples, the PROCESSES display (and summary) formats are different from those of all other classes. The PROCESSES display provides the following information:
Data Item Description PID Process identifier as assigned by the system, in hexadecimal STATE Process's scheduler state (see the description of the MONITOR STATES command for an explanation and a tabular summary of the STATES codes) PRI Current (as opposed to base) priority of the process NAME Process name PAGES Number of shareable pages and total number of pages currently in use by the process DIOCNT Cumulative direct I/O operations performed by the process since its creation; not displayed if the process is swapped out FAULTS Cumulative page faults since the process was created; not displayed if the process is swapped out CPU TIME Cumulative CPU time used by the process since its creation, in the format hours:minutes:seconds ; not displayed if the process is swapped out The top corners of the display contain the number of processes in the system and the time in days, hours, minutes, and seconds since the system was last booted. Processes that are swapped out are so noted.
If more processes are in the system than can be displayed on the terminal screen at once, the display consists of multiple screens. Screens are presented one at a time at intervals specified with the /VIEWING_TIME qualifier. The five /TOP bar graph displays provide the PID and process name of each of the top eight users.
As with the other bar graph displays, examples in the displays of top users are rounded to the nearest whole number. Up to 16 processes with nonzero values are displayed. To be eligible for inclusion in the list of top users, a process must be present and swapped in at the beginning and end of the display interval. This eligibility requirement also applies to the beginning and ending of the entire period covered by a summary.
Note that only one of the displays of top users or the regular PROCESSES display can be selected in a single MONITOR request.
| #1 | |
|---|---|
MONITOR> MONITOR/INPUT=PROCS.DAT/INTERVAL=6 PROCESSES
|
This example illustrates a PROCESSES display generated from the input file PROCS.DAT. One line is displayed for each process in the system. This display shows current values only---average, minimum, and maximum statistics are not available. Also for swapped-out processes, the words SWAPPED OUT replace the three rightmost items, because those items are not available for swapped-out processes. Because this example is a playback request, the system uptime displayed is that of the system at the time the MONITOR data was recorded.
Nondisplayable characters in process names are represented by periods.
| #2 | |
|---|---|
MONITOR> MONITOR/INPUT=PROCS.DAT PROCESSES/TOPDIO
|
This example shows that the process SAMPLE0901, with a rate of 25 per second, was the top consumer of direct I/Os during the most recent interval between displays.
| #3 | |
|---|---|
MONITOR> MONITOR PROCESSES/TOPCPU
|
This example shows a MONITOR PROCESSES/TOPCPU screen display on a single CPU system.
| #4 | |
|---|---|
MONITOR> MONITOR PROCESSES/TOPCPU
|
This example shows s MONITOR PROCESSES/TOPCPU screen display on a 12-CPU system.
| #5 | |
|---|---|
MONITOR> MONITOR PROCESSES/TOPCPU
|
This example shows s MONITOR PROCESSES/TOPCPU screen display on a 16-CPU system.
| #6 | |
|---|---|
MONITOR> MONITOR PROCESSES/TOPSUPERVISOR
|
The command in this example displays a bar graph that shows the 16 processes that are top consumers of CPU time in supervisor mode. Values are expressed in clock ticks (10 ms) per second.
The MONITOR RLOCK command initiates monitoring of the RLOCK (dynamic lock remastering) statistics class.
MONITOR RLOCK
/qualifier[,...]
One or more qualifiers as described in the Command Qualifier Descriptions section.
/ALL
Specifies that a table of all available statistics (current, average, minimum and maximum) is to be included in the display and summary outputs. For summary output, this qualifier is the default for all classes; otherwise, it is the default for all classes except CLUSTER, MODES, PROCESSES, STATES, SYSTEM, and VECTOR./AVERAGE
Specifies that a bar graph of average statistics is to be included in the display and summary outputs./CURRENT
Specifies that a bar graph of current statistics is to be included in the display and summary outputs. The /CURRENT qualifier is the default for the CLUSTER, MODES, STATES, SYSTEM, and VECTOR classes./MAXIMUM
Specifies that a bar graph of maximum statistics is to be included in the display and summary outputs./MINIMUM
Specifies that a bar graph of minimum statistics is to be included in the display and summary outputs.
You can use the RLOCK class to monitor the dynamic lock remastering statistics of a node. Because local locking operations are less costly than remote operations, lock trees are moved from node to node to improve performance. A lock tree might be moved for any of the following reasons:
- Another node in the cluster is much more active on the tree than the current master.
- A node with a higher LOCKDIRWT enqueues a lock to a resource that a node with a lower LOCKDIRWT masters.
- Only one node in the cluster has locks on this resource and should, therefore, become the master.
The class RLOCK consists of the following data items, which are displayed as the rate of occurrences per second:
Data Item Description Lock Tree Outbound Rate Rate at which lock trees are moved from this node. Higher Activity Rate for trees moved due to higher locking activity on another node in the cluster. Higher LOCKDIRWT Rate at which trees are moved to a node with a higher value of the SYSGEN parameter LOCKDIRWT. Sole Interest Rate at which trees are moved to another node because that node is the only one with locks remaining on the tree. Remaster Msg Send Rate Rate at which remaster messages are sent from this node. Lock Tree Inbound Rate Rate at which trees are moved to this node. Remaster Msg Receive Rate Rate at which remaster messages are received on this node.
MONITOR> MONITOR RLOCK
|
In this example, the outbound numbers are quite low; in most cases, these numbers are never very large. Remastering is attempted only once every 8 seconds; then a maximum of 5 trees are processed at once. The exception is during orderly shutdown, when the system attempts to force all trees off the node shutting down.
| Previous | Next | Contents | Index |