Trailing-Edge
-
PDP-10 Archives
-
bb-d549g-sb
-
sysdpy.man
There are 3 other files named sysdpy.man in the archive. Click here to see a list.
SYSDPY %434(562) users reference manual
Copyright (C) 1979
Digital Equipment Corporation, Maynard, Massachusetts, U.S.A.
This software is furnished under a license and may be used and copyed only in
accordance with the terms of such license and with the inclusion of the above
copyright notice. This software or any other copies thereof may not be provided
or otherwise made available to any other person. No title to and ownership of
the software is hereby transferred.
The information in this software is subject to change without notice and should
not be construed as a commitment by Digital Equipment Corporation.
Digital assumes no responsibility for the use or reliability of its software on
equipment which is not supplied by Digital.
SYSDPY %434(562) Users Reference Manual Page 1
Introduction
SYSDPY is a program which runs under the DECsystem-10 (TOPS-10) monitor.
It displays various job and/or system statistics on a video terminal. Unlike
SYSTAT, SYSDPY dynamically updates the video screen, changing only the
characters that have changed since the last display update. In addition, the
ability to "scroll" through a display which is larger than the screen size is
provided.
SYSDPY is available in nine different flavours (i.e., nine different
programs) to run on each of nine different types of terminals. Each different
flavour is assembled to run only on its particular type of terminal (or any
compatible terminal). The nine flavours of SYSDPY currently available are:
SYSDPA SYSDPY for a Digital VT05A terminal.
SYSDPB SYSDPY for a Digital VT05B (2400 baud VT05) terminal.
SYSDPY SYSDPY for a digital VT06 (Datapoint 3300) terminal.
SYSVBX SYSDPY for a Digital VB10C graphics display.
SYSV50 SYSDPY for a Digital VT50 terminal.
SYSV52 SYSDPY for a Digital VT52 terminal.
SYSV61 SYSDPY for a Digital VT61 terminal.
SYSDLT SYSDPY for a Delta Data Telterm terminal.
SYSHZL SYSDPY for a Hazeltine 2000 terminal.
Although no privileges are generally required to run SYSDPY in the normal
mode, certain statistics are available only by PEEKing or SPYing on the monitor,
which requires some privileges (e.g., SYSDPY running with the JACCT bit set or
under an account with PEEK or SPY privileges).
The following discussions presuppose SYSDPY running under a 7-series
monitor, but in general SYSDPY will run equally well with a 6-series monitor.
SYSDPY %434(562) Users Reference Manual Page 2
Commands
When SYSDPY is first run it will by default start the normal job display.
While SYSDPY is running it is always listening for commands.
A SYSDPY command is generally a single letter ("N", for example, selects
the normal job display which is the default on startup). No carriage return
line feed sequence is needed to terminate the command.
SYSDPY commands
A Toggle auto-roll of all jobs. Normally SYSDPY will scroll through the
entire display available, advancing by approximately one screenful each
update pass. The "A" command will toggle the current scroll or
no-scroll setting (see also the "+" and "-" commands).
F List file system statistics. The "F" command changes the display to one
which lists, for each disk unit in the system, I/O counts, error
summaries, structure membership, etc. In addition a summary of swapping
units is provided. This command requires PEEK or SPY privileges.
H List help text. If the file HLP:SYSDPY.HLP exists then read and display
the file. Otherwise a quick summary of all available commands will be
displayed.
I List incremental statistics (where applicable). The "I" command tells
SYSDPY to, rather than simply listing the usual total since system
startup value, list the incremental value, i.e., the change in the value
since the last update cycle. This incremental facility applies only to
certain values: the system uptime (except on "N" display where uptime
is always total); job runtimes; job disk reads and writes ("S"
command); all cpu statistics; NET I/O (byte) counts; TTY I/O counts;
and network message counts (the "\" command). Incremental mode is
generally indicated by a "(+)" immediately preceding the uptime value
(except on the "N" display where it immediately precedes the first cpu
column header). The "J" display and the "N" display in "S" mode do not
display the incremental mode indicator.
J List only jobs (in multiple columns). The "J" command is similar to the
"N" command but restricted to displaying only jobs (no system
statistics). As many columns of jobs will be displayed as will fit on
the screen (each column requres 40 spaces).
L Log screens into file LPT:SYSDPY.LOG[-] (close on ^Z). The "L" command
instructs SYSDPY to, each time the screen is updated, write the entire
contents of the screen into the "log" file. If the file already exists
the new data is appended to the already extant file. Each screen is
separated with a form feed. The file is CLOSEd on a "^Z" command to
SYSDPY.
M List memory data. The "M" command displays the normal job data but with
an expanded memory summary showing both virtual and physical job memory
usage. No system statistics are displayed in the "M" job display.
N List normal (some of everything) status. The "N" command lists a column
of job status information (same as "J" display) and a column of system
status information such as cpu idle and lost time, available disk
SYSDPY %434(562) Users Reference Manual Page 3
Commands
structures, etc. This is the default state when SYSDPY is first
started.
O Toggle suppression of [OPR] jobs in "J", "N", and "M" displays.
Normally SYSDPY will display all jobs under all ppns. The "O" command
will instruct SYSDPY to suppress display of jobs running under the [1,2]
([OPR]) ppn. A second "O" command "toggles" back to displaying [OPR]
jobs.
Q List the system queues. The "Q" command displays the system's queues
just as the QUEUE program would, but in "DPY" mode. This command
requires GALAXY version 4.
R Refresh entire screen immediately. The "R" command forces an immediate
screen refresh cycle (a refresh cycle re-writes the entire screen rather
than just updating only the portions of the screen that changed). By
default, SYSDPY will automatically refresh the screen every so often
anyway (every 100 update cycles for VT06's, 1000 update cycles for
VT05's, or 10000 update cycles for any other type of terminal
supported). The "R" command resets the auto-refresh counter. All
screen refreshes are made in normal video regardless of the state of the
reverse video update mode ("^R" command).
S Toggle system or expanded job statistics in "N" display. The "S"
command will cause SYSDPY to display the disk reads, disk writes, and
the user name for jobs in the "N" display rather than the usual system
statistics. A second "S" command toggles back to showing the system
statistics.
T Topology for network display. The "T" command changes the display to
show the network topology and message numbers (NOT the number of
messages but the NCL message numbers). This display requires PEEK or
SPY privileges.
V Toggle cpu and core priority. The "V" command requests SYSDPY to
attempt to lock in memory and run in a high priority run queue. If
SYSDPY was entered in an HPQ state (a .SET HPQ command is in effect for
the job) then SYSDPY will not change the HPQ setting. In addition, if
no "W" command has been issued the update time is reduced to one second.
A second "V" command will clear the lock and HPQ setting (except that if
a .SET HPQ command is in effect HPQ is not cleared) and restore the
update time to the default ten seconds (if no "W" command has been
issued). After 1000 update cycles SYSDPY automatically clears the "V"
state. No error message is issued if SYSDPY fails either to lock or run
in HPQ. This command requires LOCK and/or HPQ privileges.
nW Set wait time to "n" decimal seconds. The "W" command is used to select
the wait time between screen updates. The default update time is ten
seconds (one second if in "V" mode). The update interval may range from
zero to sixty seconds. Note that a value of zero means one clock tick,
not infinite run state. If "n" is not typed SYSDPY will revert to the
default wait time.
\ Display network statistics. The "\" command selects the network
statistics display. The number of network messages sent and received is
SYSDPY %434(562) Users Reference Manual Page 4
Commands
shown, broken down by message type (e.g., DATA or NEIGHBORS). In
addition, a histogram is shown of the data messages sent and received as
a function (log base 2) of the data message size. Also, network free
core usage is shown.
n+ Advance screen by "n" or approximately one screen-full if "n" is not
specified. The "+" command scrolls the screen forward by "n" lines or
jobs (depending on the display) or, if "n" is not specified, by slightly
less than a full screen. If the end of the display is reached SYSDPY
will "bottom" justify the display to put as much of the display as
possible onto the screen. The issuance of the "+" command clears
auto-roll (see the "A" command).
n- Retreat screen by "n" or approximately one screen-full if "n" is not
specified. The "-" command scrolls the screen backwards by "n" lines or
jobs (depending on the display) or, if "n" is not specified, by slightly
less than a full screen. If the end of the display is reached SYSDPY
will "top" justify the display to put as much of the display as possible
onto the screen. The issuance of the "-" command clears auto-roll (see
the "A" command).
# Toggle PTY number or controlling job number. The "#" command toggles
whether the "Where" column (of the job display) shows the PTY number or
the controlling job number for a job which is being controlled by
another job.
% Toggle runtime or percentage of cpu. The "%" command toggles whether
the "Runtime" column (of the job display) shows the actual runtime in
hours, minutes, seconds, and hundredths, or the percentage of the cpu
that the runtime represents. NOTE: on KL-10 based systems which
calculate the user program runtime from EBOX/MBOX ticks the runtime
figure can vary tremendously depending on the cache hit rate. For a
totally compute bound job (no monitor calls) the runtime can vary from
approximately 12% of elapsed time with a 0% cache hit ratio to over 110%
of elapsed time with a 100% cache hit ratio.
^A Toggle alarm (flashing) mode (if applicable). The "^A" command causes
the entire screen to enter alarm mode (VT61's only). A second "^A"
command clears alarm mode. SYSDPY will also clear alarm mode on exit.
^C Terminate SYSDPY execution. The "^C" command will cause SYSDPY to exit
at the end of the current update pass if one is in progress, or
immediately if an update is not in progress. SYSDPY will close any log
file being written, clear alarm and reverse video modes, and position
the cursor to the bottom of the screen before exiting.
^R Toggle reverse video updating (if applicable). The "^R" command causes
all subsequent screen updates to be made in reverse video (VT61's only).
A second "^R" command clears reverse video update mode. SYSDPY will
also clear reverse video update mode on exit. All screen refreshes
(always implied when a new display is started) are made in normal video
regardless of the setting of the reverse video update mode.
^Z Terminate SYSDPY execution. The "^Z" command will cause SYSDPY to exit
at the end of the current update pass if one is in progress, or
SYSDPY %434(562) Users Reference Manual Page 5
Commands
immediately if an update is not in progress. SYSDPY will close any log
file being written, clear alarm and reverse video modes, and position
the cursor to the bottom of the screen before exiting.
<ESC> Freeze screen as is (any subsequent command will thaw). The <ESC>
command will cause SYSDPY to cease updating the screen until another
command is typed. Any command typed (except of course another <ESC>
command) will enable SYSDPY to resume screen updating.
SPACE Update screen display immediately. SYSDPY will always update the screen
display after issuance of any command (except <ESC>). SPACE will force
an update cycle without changing any SYSDPY parameters.
SYSDPY %434(562) Users Reference Manual Page 6
F display - file system statistics
Following is a sample of the SYSDPY "F" display:
This is RD352A KL10 SYS#1279 28-May-79 23:19:00 UP:14:27:58
Unit or F/S Free BR BW DR DW MR MW Seeks
Structure DSKB Free:9388 Mount:15
RPA0(DSKB0) 9388 38481 9735 26491 13255 24359 10334 27237
Errors: SDAT:4 RETRIES:1 LBN:147276 1CONI:500000,202414 2CONI:500000,202414
1DATAI:100000,51700 2DATAI:100000,51700
RPB0(.....) 12201 9883 4005 3914 2208 1784 5090
Structure DSKH Free:6145 Mount:15
DPA0(DSKH0) 6145 810 1011 302 1 726 145 542
Structure RETI Free:116 Mount:0
DPA1(KALMAN) 116 239 0 827 0 567 9 759
Structure KR Free:12485 Mount:0
DPA2(DAS001) 12485 4 2 350 0 506 14 324
Swap Unit R W Used(P)
RPA0(DSKB0) 39351 26800 495/2048 = 24%
RPB0(.....) 2354 1001
DPA0(DSKH0) 0 0 0/1024 = 0%
The first line is the standard SYSDPY "ID" line, containing the name of the
system (preceded by the header "This is"), the current date and time, and
finally the system uptime.
The "F" display is divided into two sub-displays: disk file structure and
unit information, and swapping unit information. The first portion of the
display is devoted to disk file structures. Each structure is first identified:
Structure The logical name of the disk file structure.
Free The total free blocks for the entire structure.
Mount The total mount count for the structure (note that being in the active
swapping list counts as a "user", as does being in the system search
list).
Next, each disk unit in the file structure is summarized. First the count
of I/O operations for that unit is listed in columns as follows:
Unit or F/S This column identifies first the physical disk unit or drive and
second the logical disk unit ID (enclosed within parenthesis) within a
file structure which is mounted on that physical drive. A disk unit ID
of "....." indicates that this physical unit is an alternate port for
the immediately preceding disk unit or drive.
Free Free count. The number of free blocks available on the specific unit
(as opposed to the entire file structure). For alternate ports of a
multi-ported disk the free count is blank (since the free count is
dependent on the pack mounted, not the access port).
BR Buffered Reads. The count of user-mode buffered I/O reads (i.e., disk
reads done in ASCII, ASCII line, byte, image, image binary, or binary
I/O mode) for that unit in disk blocks.
BW Buffered Writes. The count of user-mode buffered I/O writes (i.e., disk
writes done in ASCII, ASCII line, byte, image, image binary, or binary
I/O mode) for that unit in disk blocks.
DR Dump Reads. The count of user-mode dump I/O reads (i.e., disk reads
SYSDPY %434(562) Users Reference Manual Page 7
F display - file system statistics
done in image dump, dump records, or dump record I/O mode) for that unit
in disk blocks.
DW Dump Writes. The count of user-mode dump I/O writes (i.e., disk writes
done in image dump, dump records, or dump record I/O mode) for that unit
in disk blocks.
MR Monitor Reads. The count of monitor reads (i.e., disk blocks read by
the monitor on behalf of a user, as in reading directories or RIBs) for
that unit in disk blocks.
MW Monitor Writes. The count of monitor writes (i.e., disk blocks written
by the monitor on behalf of a user, as in updating SATs) for that unit
in disk blocks.
Seeks Seek count. The count of positioning operations performed by the
monitor for that unit.
Following the I/O operation counts for each unit, the status of each unit
is detailed as follows:
RHB Re-read Home Blocks. The unit has gone offline or unsafe and the
monitor must read the disk pack home blocks again in order to verify
that the correct disk pack is still mounted on that unit
OFL OFfLine. The unit is currently offline (although the monitor thinks
there should be a disk mounted and accessible on that unit).
HWP Hardware Write Protect. The unit is write protected by hardware rather
than software. The monitor is totally unable to write onto disk unit
(e.g., for swapping).
SWP Software Write Protect. The unit is write protected by software for
this job. The monitor is able to write onto the disk unit (e.g., for
updating file access dates), and other jobs are able to write onto the
unit.
SAF Single Access File structure. The unit is part of a single-access file
structure (i.e., only one job is allowed to access the disk unit).
PRF PRivate File structure. The unit is part of a "private" file structure.
A private file structure is one that can be mounted only by its owner's
project group or by a privileged job.
NNA No New Accesses. The unit is "locked" by the operator (i.e., no new
file accesses by users are allowed). Typically this is done prior to
the removal of a file structure in order to let file accesses "wind
down" and gracefully terminate before the file structure is actually
removed from the system.
AWL Write Locked for All. The unit is write locked for all jobs in the
system (unlike SWP above which applies only to this job).
HDEV Hard DEVice errors. This is the total count of non-recoverable errors
other than data errors (i.e., the hardware is at fault (such as a power
supply failure), not the data).
HDAT Hard DATa errors. This is the total count of non-recoverable errors in
the data itself (i.e., the hardware is functioning correctly but the
data (or data media) is at fault).
SDEV Soft DEVice errors. This is the total count of recoverable errors other
than data errors.
SDAT Soft DATa errors. This is the total count of recoverable errors in the
data itself.
RETRIES The number of times the monitor tried the last operation before it
finally succeeded (soft error condition) or failed (hard error
condition).
SER SAT ERrors. This is the count of the number of times that a disk SAT
SYSDPY %434(562) Users Reference Manual Page 8
F display - file system statistics
block disagreed with the monitor's main memory count of the number of
free blocks on the disk unit.
RER RIB ERrors. This is the count of RIB (Retrieval Information Block)
errors encountered in files on the disk unit.
CER Checksum ERrors. This is the count of checksum errors encountered in
files on the disk unit. A checksum error occurs when the checksum field
in a RIB pointer does not match the calculated checksum of the first
word pointed to by the RIB pointer.
PHUNG Position HUNG count. This is the count of the number of times the disk
unit "hung" during a positioning operation.
THUNG Transfer HUNG count. This is the count of the number of times the disk
unit "hung" during a transfer operation but the monitor managed to
recover (i.e., get the unit going again and complete the transfer
operation) by stopping the unit in transfer state.
NTHUNG Non-recoverable Transfer HUNG count. This is the count of the number of
times the disk unit "hung" during a transfer operation and the monitor
was unable to recover and complete the transfer.
SHUNG Software HUNG count. This is the count of the number of times the
monitor itself got confused (e.g., disk unit and channel idle but in
transfer wait state) but noticed and recovered anyway.
LBN Logical Block Number. This is the logical block number (within the disk
unit) for which the last error occurred.
1CONI This is the disk unit status (CONI) after the last initial error
occurred and before any recovery was attempted.
2CONI This is the disk unit status (CONI) after the last soft error recovery
or hard error fatality.
1DATAI This is the disk unit status (DATAI) after the last initial error
occurred and before and recovery was attempted.
2DATAI This is the disk unit status (DATAI) after the last soft error recovery
or hard error fatality.
After all file structures have been listed a summary is given of the
swapping units as follows:
Swap Unit As in the "Unit or F/S" column above this column identifies both
the physical and logical disk unit.
R The total count of swap reads (in blocks) for this unit.
W The total count of swap writes (in blocks) for this unit.
Used (P) The amount of space currently in use on the swapping unit and
the total swapping space available on that unit (separated by a "/").
Note that this includes any dormant high segments residing on that unit
(which is not included in the SWP field of the "N" display). This field
is left blank for alternate ports as, like the free count above, the
space used is dependent on the pack itself and not on the ports.
Finally, if there have been any swapping I/O errors a line is printed
summarizing the errors:
Swap errors The total number of swap I/O errors.
CHK At least one checksum error occured.
DEV At least one device error occured.
DAT At least one data error occured.
Lost Total swapping space lost due to I/O errors.
SYSDPY %434(562) Users Reference Manual Page 9
M display - job and memory statistics
Following is a sample of the SYSDPY "M" display:
This is RW340A KL10 SYS#1279 01-Mar-79 04:13:02 UP:00:33:00
VM(MAX=2596,USED=476) PGR.USER=10.0 PFC:NIW=8 IW=1
Job Who Where What Virtual Physical PGR State Runtime
1 [OPR] CTY OPSER 2+6(512) 2+6+1(512) 0.00 HB@S 00:00:00
2 [OPR] D73 DAEMON 18+0(512) 18+0+2(512) 0.00 HS+S 00:00:00
3 [OPR] D74 FILDAE 27+0(512) 27+0+1(512) 0.00 SL S 00:00:00
4 [OPR] D73 MIC 2+12(512) 2+12+2(512) 0.00 HS 00:00:00
5 [OPR] J1 SYSINF 12+0(512) 12+0+2(512) 0.00 HB S 00:00:00
6 [OPR] J1 1+0(512) 1+0+2(512) 0.00 ^C S 00:00:00
7 [OPR] J1 QUASAR 14+0(512) 14+0+2(512) 0.00 HB S 00:00:00
8 [OPR] J1 BATCON 4+7(512) 4+7+1(512) 0.00 HB S 00:00:00
9 [OPR] J1 LPTSPL 9+7(512) 9+7+1(512) 0.00 TI S 00:00:00
10 [OPR] J1 NZDSPL 25+8(512) 25+8+1(512) 0.00 TI 00:00:00
11 [OPR] J1 NZDSPL 25+8(512) 25+8+1(512) 0.00 TI 00:00:00
12 [OPR] J1 JUSTIF 31+30(512) 31+30+1(512) 0.00 HS S 00:00:00
13 [OPR] J1 OPROMO 6+12(512) 6+12+2(512) 0.00 HS S 00:00:00
14 [OPR] J1 NIK 4+15(512) 4+15+1(512) 0.00 EN S 00:00:00
15 [SELF] 12 SYSV52 14+0(512) 14+0+2(512) 0.00 RN+ 00:00:18
16 226,4563 13 DSKIO 30+10(512) 30+10+9(512) 0.00 RNO 00:00:53
17 226,4563 14 MCS-10 433+46(512) 31+46+2(512) 10.81 HS V 00:00:01
18 226,4563 15 TTYTST 1+2(512) 1+2+2(512) 0.00 TON 00:00:27
19 226,4563 16 UUOS 23+1(512) 23+1+2(512) 0.00 RN# 00:00:33
20 226,4563 D16 NIK 6+15(512) 6+15+2(512) 0.00 RN 00:00:19
The first line is the standard SYSDPY "ID" line, containing the name of the
system (preceded by the header "This is"), the current date and time, and
finally the system uptime.
The second line summarizes general virtual memory usage as follows:
VM System-wide virtual memory usage as follows:
MAX The amount of virtual memory still available for all users
(i.e., total swapping space available minus the amount of
virtual memory currently being used).
USED The amount of virtual memory currently in use by all jobs.
PGR.USER The total system-wide paging rate for all users.
PFC Page Fault Class. This is the class of page fault which caused paging
operations.
NIW Not In Working set. This class of page fault occurs for pages
which are not in the job's working set, i.e., those pages which
are paged out to the swapping media (this includes allocated but
zero pages).
IW In Working set. This class of page faults occurs for pages
which are in the job's working set but which have access allowed
turned off, i.e., those pages which do not require being read in
from the swapping space.
The remainder of the screen is devoted to job display.
SYSDPY %434(562) Users Reference Manual Page 10
M display - job and memory statistics
The "Job", "Who", "Where", "What", "State", and "Runtime" columns are the
same as for the "N" display.
The memory statistics are broken down into two major columns (each with
several fields) to describe the job's virtual and physical (working set) memory
usage as follows:
Virtual This column first lists the job's low segment virtual size, followed by
the job's high segment virtual size, followed by the job's virtual
memory limit (enclosed in parenthesis).
Physical This column lists first the job's low segment physical size,
followed by the job's high segment physical size, followed by the job's
"monitor's per-process" size (which includes the job's page map, disk
DDBs, TMPCOR, logical name definitions, and so on), and finally the
job's physical memory limit (enclosed in parenthesis). The first three
fields comprise the job's working set.
SYSDPY %434(562) Users Reference Manual Page 11
N display - job and system statistics
Following is a sample of the SYSDPY "N" display:
Job Who Where What #P State Runtime This is RW341A KL10 SYS#1279
1 [OPR] CTY OPSER 9 HB@S 00:00.00 09-Mar-79 29:44.21 UP:05:49:41
2 [OPR] D73 DAEMON 20 HS+S 00:00.00 (+) ID OV LS UPTIME CTX UUO CSH
3 [OPR] D74 FILDAE 28 SL S 00:00.00 CPU0 13 26 0 00:07.50 110 426 17
4 [OPR] D73 MIC 16 HS 00:00.00 DSKI DSKO SWPI SWPO
5 [OPR] J1 SYSINF 14 HB S 00:00.00 CPU0 247 103 11 18
6 [OPR] J1 3 ^C S 00:00.00 CH0 CH1 CH2 CH3 CH4 CH5 CH6 CH7
7 [OPR] J1 QUASAR 16 HB 00:00.00 CPU0 15. .00 .00 .00 .00 .00 .00 .00
8 [OPR] J1 BATCON 12 HB S 00:00.00 PI0 PI1 PI2 PI3 PI4 PI5 PI6 PI7
9 [OPR] J1 LPTSPL 17 TI S 00:00.00 CPU0 .00 .00 .46 .56 5.5 7.5 .00 5.6
10 [OPR] J1 NZDSPL 34 TI S 00:00.00 NET In:93 Out:1058 Cor:3403
11 [OPR] J1 NZDSPL 34 TI S 00:00.00 TTY In:0 Out:905 Cnk:74%
12 [OPR] J1 JUSTIF 62 HS S 00:00.00 Mem:250/252/281 Shr:74 JRN:3/5/22
13 [OPR] J1 OPROMO 20 HS 00:00.00 Use:709/1111 Swp:774/3072 ASR:0.0
14 [OPR] J1 NIK 20 EN S 00:00.00 HDE:5 SCH:400
15 [OPR] 12 BACKUP 29 DI 00:00.00 Jobs:23/40 Log'd in:23 Det:3
16 [SELF] 13 SYSV61 29 H1&+ 00:00.10 Struc Mnt Free Struc Mnt Free
17 226,4563 14 PIP 42 DO 00:01.86 DSKB 24 10172 DSKH 9 17617
18 226,4563 15 UUOS 25 RN# 00:00.23 Dev By How Dev By How Dev By How
19 226,4563 16 TTYTST 5 TON 00:00.36 MPX22 20 I MPX21 20 I MPX20 20 I
20 226,4563 17 MCS-10 81 HS V 00:00.00 MTA100 15 AI RDA517 20 I RDA516 20 I
21 226,4563 J20 SIGNON 42 SL S 00:00.00 RDA515 20 I RDA514 20 I RDA513 20 I
22 226,4563 J20 MPPGLX 92 SL S 00:00.00 RDA512 20 I RDA511 20 I RDA510 20 I
23 226,4563 J20 MPPSPL 59 SL S 00:00.00
This display has two completely independent componets - a job display (the
left half of the screen) and the system display (the right half of the screen).
If "S" mode is in effect then the system display is not shown (the job display
is expanded to fill the screen).
The job display shows all the jobs in use, by whom, doing what, etc. The
job information displayed is as follows:
Job The job number.
Who The user account running in that job slot. Normally the user ppn is
listed. There are two special cases:
[OPR] The "user" is the privileged operator account (usually [1,2]).
[SELF] The "user" is running under the same account under which SYSDPY
is running.
Where The terminal which is controlling the job, if any. This field has
several possible representations:
CTY The terminal is the system CTY.
DET There is no terminal attached to that job. If the job has PEEK
or SPY privileges this will be expanded to include the terminal
to which the job was last attached.
Dnn There is no terminal currently attached to that job. The job
SYSDPY %434(562) Users Reference Manual Page 12
N display - job and system statistics
was last attached to TTYnn. This field requires PEEK or SPY
privileges.
Jnn The job is being controlled by job nn (i.e., the job is attached
to a PTY owned by job nn). This is the normal display for a
controlled job but it may be changed by the "#" command to
display the PTY number instead.
Pnn The job is attached to PTYnn (i.e., the job is being controlled
by another job). The "#" command may be used to display the
controlling job number instead.
nn The job is attached to TTYnn.
What The job's logical program name. Although usually the name of the
program last run by that job, the job may have issued a SETNAM monitor
call to change the program name.
#P The total user working set size (including "monitor per process" pages
such as the page map).
State The state of the job. This column consists of the basic state code (two
letters and/or digits) plus assorted flags. The basic state code
(running, in I/O wait, etc.) is shown first. Although nominally the
state code names are the same as the names of the actual monitor job
queues some of the state names are "SYSDPY Specials" which are intended
to convey more information about the job than just the job queue in
which the job currently resides:
AU Alter UFD (or SFD) wait. The AU "resource" is essentially the
monitors file directory interlock. In order to change a file
directory (e.g., create or delete files) the job must have the
file directory interlock. The directory interlock is on a per
directory per structure basis. Note that this file directory
interlock is NOT the same as the STRUUO UFD interlock.
CA Core Allocation wait. The job is attempting to lock in
physically contiguous memory but must wait for the monitor to
move (swap) other jobs out of the way first.
CB Core Block wait. The CB "resource" is essentially the monitor's
file system interlock based on the monitor's file system
database in monitor free core - hence the name "Core Block".
Only one job may have the CB resource at any time, all other
jobs must queue up for it to become available.
CW Command Wait. Although the user program is still running (i.e.,
it is not in ^C state) the user typed a command which is still
pending on the job's controlling terminal. This is a SYSDPY
Special state, the job can be in any of the queues except the ST
queue.
DA Disk Allocation wait. The DA "resource" is essentially the
monitor's disk unit SAT (Storage Allocation Table) interlock.
Only one job may have the DA resource for any specific disk unit
at any time, all other jobs must queue up for it to become
available.
DC Data Control wait.
DI Disk Input wait. The job is blocked waiting its turn to read
from a disk unit. This is a SYSDPY Special state which requires
PEEK or SPY privileges, the job is actually in the DI queue.
DO Disk Output wait. The job is blocked waiting its turn to write
to a disk unit. This is a SYSDPY Special state which requires
PEEK or SPY privileges, the job is actually in the DI queue.
SYSDPY %434(562) Users Reference Manual Page 13
N display - job and system statistics
DS Disk I/O wait Satisfied. Formerly in DI/DO/DW state, the job's
disk I/O request has been completed at disk interrupt level.
The job is waiting for the scheduler to requeue the job into the
run queue(s).
DW Disk I/O wait. The job is blocked waiting its turn to either
read or write form or to a disk unit. This is a SYSDPY Special
state, the job is actually in the DI queue.
D1 DECtape controller number 1 wait. The job is waiting for the
first DECtape controller to become available for some sort of
I/O operation.
D2 DECtape controller number 2 wait. The job is waiting for for
the second DECtape controller to become available for some sort
of I/O operation.
E6 Event DN60 wait. The job is waiting for a DN60-class front end
operation to complete. This is a SYSDPY Special state, the job
is actually in the EW queue.
EF Event Front-end wait. The job is performing some sort of
operation involving a "front-end" processor and is waiting for
the operation to complete. This is a SYSDPY Special state, the
job is actually in the EW queue.
EK Event Kontroller wait. The job is requesting the monitor to
perform some magnetic tape operation which is blocked pending
the availability of the tape kontroller. This is a SYSDPY
Special state, the job is actually in the EW queue.
EL Event Label wait. The job is performing some sort of magnetic
tape operation which requires the intervention of the tape label
processor. This is a SYSDPY Special state, the job is actually
in the EW queue.
EN Event Network wait. The job is performing some sort of
operation with a network device which is blocked pending
completion. This is a SYSPDY Special state, the job is actually
in the EW queue.
ER Event Rewind wait. The job is waiting for a magnetic tape
rewind operation. This is a SYSDPY Special state, the job is
actually in the EW queue.
ES Event wait Satisfied. Formerly in some sort of event wait, the
job is now ready to run again, and is waiting for the monitor to
reschedule the job into the run queues. This is a SYSDPY
Special state, the job is actually in the EW queue(s).
EV EVM (Exec Virtual Memory) wait. The job is attempting some
(probably I/O) operation which requires some part of the job's
address space to be mapped into the monitor's address space,
(such as DECtape I/O). The job must wait until other jobs free
up sufficient EVM for the monitor to honor the job's request.
EW Event Wait. This wait state covers a wide range of conditions,
such as waiting for a magnetic tape to rewind or a network task
channel connection. This is sort of a catch-all wait state for
any type of wait condition which does not deserve a monitor
queue all to itself (such as the "DI/DO/DS" state above).
Hn The job is in the run queue(s) and is running in high priority
run queue "n". This is a SYSDPY Special state, the job is
actually in the RN queue(s).
HB HiBernate. The job has suspended execution pending some sort of
external awakening condition, such as I/O completion or
reception of an IPCF packet (a HIBER with a zero (infinite)
SYSDPY %434(562) Users Reference Manual Page 14
N display - job and system statistics
SLEEP time). This is a SYSDPY Special state, the job is
actually in the SL queue.
HS Hibernate with Sleep. The job has suspended execution pending
either some sort of external awakening condition or the passage
of a specified time interval (a HIBER with a non-zero SLEEP
time). This is a SYSDPY Special state, the job is actually in
the SL queue.
IN INput wait. The job is waiting for some sort of input operation
to complete. This is a SYSDPY Special state which requires PEEK
or SPY privileges, the job is actually in the IO queue.
IO I/O wait. The job is waiting for some sort of currently active
I/O operation to complete, or possibly the monitor has stopped
the job and is waiting for the job's I/O to complete so the job
can be swapped.
JD Job DAEMON wait. The job has issued a DAEMON monitor call but
DAEMON has not finished servicing the request yet.
MM Memory Management wait. The MM "resource" is essentially the
monitor's main memory interlock. A program must "own" the MM
resource before it can change its main memory image (e.g.,
allocate more memory via a CORE or PAGE. monitor call). Only
one job may own the MM resource, all other jobs must queue up
for it to become available.
NA NAp. The job has temporarily suspended active execution (SLEEP
or HIBER) for a very short interval (less than one second).
NU NUll. The NU queue contains all the jobs which are not in use.
Since SYSDPY does not display jobs which are not in use this
state should never appear.
OU OUtput wait. The job is waiting for some sort of output
operation to complete. This is a SYSDPY special state which
requires PEEK or SPY privileges, the job is actually in the IO
queue.
OW Operator Wait. The job is in a ^C state due to a device error,
and is waiting to be continued (by the operator). This is a
SYSDPY special state, the job is actually in the ST queue.
PI Paging I/O wait. The job has requested the monitor to change
the job's working set in a manner which requires either reading
or writing pages to or from the swapping space.
PS Paging I/O wait Satisfied. Formerly in PI state, the job's
paging I/O request has been completed at disk interrupt level.
The job is waiting for the scheduler to requeue the job into the
run queue(s).
RU RUN or GETSEG or MERGE in progress. The job is in the process
of performing a GET command, a GETSEG monitor call, a MERGE
command or monitor call, or a RUN (or R) command or monitor
call. This is a SYSDPY special state, the job can actually be
in one of several different job queues.
RN RuN. The job is in the run queue(s), either actively running or
awaiting its turn to run on some cpu.
SL SLeep. The job has temporarily suspended active execution
(SLEEP) for a fairly long (one second or more) interval. This
is a SYSDPY Special state, the job is actually in the SL queue
(see the "HS" state above).
ST STopped. The job is HALTed. The program may have executed a
HALT instruction or EXIT monitor call, the user may have
manually stopped the program (^C or .HALT monitor command), or
SYSDPY %434(562) Users Reference Manual Page 15
N display - job and system statistics
the monitor may have stopped the program due to some error
condition. SYSDPY will never display this state but will
attempt to SYSDPY Special the state into ^C, ^W, OW, etc.,
states.
TI Terminal Input wait. The job is blocked waiting for input from
a terminal (usually the job's controlling or command terminal).
This is a SYSDPY Special state which requires PEEK or SPY
privileges, the job is actually in the TI queue (see the "TO"
state below).
TO Terminal Output wait. The job is blocked waiting for terminal
output to complete (actually, since all terminal I/O is buffered
through the monitor the job is waiting for the terminal to
accept sufficient output that the monitor can finish copying the
job's terminal output into the monitor's internal buffers).
This is a SYSDPY Special state which requires PEEK or SPY
privileges, the job is actually in the TI queue (see the "TI"
state above).
TS Terminal I/O wait Satisfied. Formerly in either TI or TO state,
the job's terminal I/O request has been completed at terminal
interrupt level. The job is waiting for the scheduler to
requeue the job into the run queue(s).
TW Terminal I/O Wait. The job is blocked waiting for terminal
input or output to complete. This is a SYSDPY Special state,
the job is actually in the TI queue.
WS Wait Satisfied. Formerly in an I/O wait state, the job's I/O
request has been completed at interrupt level. The job is
waiting for the scheduler to requeue the job into the run
queue(s).
^C The job is HALTed and completely idle. This is a SYSDPY Special
state, the job is actually in the ST queue.
^D The job is HALTed but a DAEMON request is being serviced for the
job (e.g., a "DCORE" command). This is a SYSDPY Special state,
the job is actually in the ST queue.
^W The job is HALTed but waiting for a command to be processed
(such as an "E" command but the program is swapped out). This
is a SYSDPY Special state, the job is actually in the ST queue.
nn The job is in the run queue(s) and is running in high priority
run queue "nn". This is a SYSDPY Special state, the job is
actually in the RN queue(s).
After the basic job state is shown the following flags may appear:
N The job's high segment is a sharable high segment which comes
from the NEW: device (usually [1,5]).
O The job's high segment is a sharable high segment which comes
from the OLD: device (usually [1,3]).
# The job's high segment is a sharable high segment which comes
from a non-system directory (i.e., neither OLD:, STD:, or NEW:).
@ The job's high segment has been obsoleted (i.e., the disk file
from which the high segment originated has been deleted or
superceded).
+ The job's high segment is actually a SPY segment.
& The job's low segment is locked in physical memory.
S The job's working set is swapped out to secondary storage.
F The job's working set is swapped out to secondary storage in a
SYSDPY %434(562) Users Reference Manual Page 16
N display - job and system statistics
non-contiguous fashion - i.e., the job's working set is swapped
and fragmented. If not preceded by an "S" (indicating a swapped
working set) then the working set is in main memory but the
swapping copy of the high segment is fragmented.
V The job is Virtual (i.e., the job's working set is only a part
of the job's total virtual memory image).
Runtime The job's program execution time in one of two forms: if the runtime is
at least one hour then in the form hh:mm:ss (hours, minutes, and
seconds); if the runtime is less than an hour then in the form mm:ss.cc
(minutes, seconds, and hundredths). If the "%" command is in effect
than the runtime itself will be converted to a percentage of cpu figure
instead of a straight elapsed time value.
Reads The total disk reads (in blocks) performed by the job. This column is
available only in "S" mode.
Writes The total disk writes (in blocks) performed by the job. This column is
available only in "S" mode.
User The user name.
The system display is as follows:
The first line identifies the name of the monitor currently running.
The second line lists the current date, time, and system uptime.
Next is the first of four per-cpu statistics blocks as follows:
ID Idle time (percent). The idle time is the time during which the cpu has
no user job which wants to run.
OV Overhead time (percent). The overhead time is the time spent doing
monitor processing which cannot be readily attributed to a specific job
(e.g., scheduling jobs or servicing interrupts).
LS Lost time (percent). The lost time is the time that the monitor wanted
to run a user job but couldn't because (e.g.,) another job had to be
swapped out in order to make room to swap in the runnable user job AND
there were no other user jobs runnable in the interim.
UPTIME Uptime (either hh:mm:ss or mm:ss.cc). The cpu uptime is how long the
cpu has been running. Note that this figure is completely independent
of the system uptime.
CTX Context switches per second. A context switch is switching from one
user job to run another user job.
UUO Monitor calls per second.
DBL Cpu doorbells per second (multi-cpu systems only). A cpu doorbell is a
notification from another cpu that something has happened which might
have caused a job to become runnable on this cpu. A doorbell counts
only if the cpu is idle (i.e., running the null job).
CSH Cache sweeps per second (KL-10 cpus only).
Following the general cpu statistics block is a disk I/O statistics block.
This block summarizes the total disk I/O which has passed through each cpu:
DSKI DiSK Inputs. This is the total number of disk blocks read (other than
swapping) per second by this cpu.
SYSDPY %434(562) Users Reference Manual Page 17
N display - job and system statistics
DSKO DiSK Outputs. This is the total number of disk blocks written (other
than swapping) per second by this cpu.
SWPI SWaP Inputs. This is the total number of disk blocks read for swapping
per second by this cpu.
SWPO SWaP Outputs. This is the total number of disk blocks written for
swapping per second by this cpu.
Following the cpu disk statistics block is the internal channel and PI
usage block. This block is available only on KL-10 processors, and only if the
monitor's "background performance analysis" facility is turned on (PERF.
function .PMBPN).
The first set of figures are the RH20 internal channel usages, in percent
(11. is eleven percent, 1.1 is one and one one-tenth percent, .11 is
eleven-hundredths of one percent). These percentages represent the time that
the RH20 has data for the MBOX, as opposed to the time the RH20 is actually
busy. As such the values can vary widely (easily a factor of ten to one) for a
given RH20 usage - the heavier the system (MBOX) load (other RH20s, cpu cache
sweeps, etc.) the higher the channel percentages will be.
The second set of figures are the PI usages, again in percent (as above).
PI0 is a DTE20 cycle-stealing from the cpu. In the example above, PI3 is
SCNSER, PI4 is the disk service, PI5 is NETSER (shipping the TTY characters to
the DN87), and PI7 is the scheduler.
Following the cpu statistics blocks is the network I/O count summary:
In Network bytes received and processed per second.
Out Network bytes processed and transmitted per second.
Cor Amount of monitor free core in use by NETSER.
Next is a TTY I/O count summary:
In TTY characters received (typed by users, not MIC) per second.
Out TTY characters output (by programs or echoed, includes fill) per second.
Cnk TTY chunks used (percent).
Next are the memory and job loading statistics:
Mem The amount of physical system memory available for all users. There are
three fields: maximum single user (CORMAX); maximum for all users
combined (MAXMAX); and the maximum possible MAXMAX (PHYMAX), i.e., the
amount of memory for which the system is configured. If all three
values are the same, only CORMAX is typed. If MAXMAX and PHYMAX are the
same but CORMAX is different then only CORMAX and MAXMAX are typed. If
CORMAX and MAXMAX are the same but PHYMAX is different (i.e., some
memory is locked by user programs or is offline) then only CORMAX and
PHYMAX are typed (MAXMAX is left blank). If all three are different
then all three are typed.
Shr The amount of virtual memory saved by sharing high segments.
JRN The number of Jobs RuNning. There are three fields: the number of
running jobs (i.e., in the RN queue(s)), the number of jobs running or
I/O blocked (i.e., neither stopped nor one of SL or HB or TI or EW
states), and the number of jobs doing anything (i.e., not stopped).
Use The total user memory space used by all users. There are two fields:
SYSDPY %434(562) Users Reference Manual Page 18
N display - job and system statistics
working set user memory, and virtual user memory. Only if the two are
different are both fields typed.
Swp The total swapping space used. There are two fields: total user space
used by all jobs, and total space available to all jobs.
ASR Active Swapping Ratio. The ratio of total active user working set
memory space (active defined as second field of JRN above) minus MAXMAX
to MAXMAX (i.e., the ratio of how much is swapped out to how much room
is available for it in physical system memory).
Next is a line of system shutdown timer, current scheduling, and error
counts as follows:
KSYS Minutes remaining until system shutdown.
HDE Total hardware error count.
STP Stopcode count in two fields: job stopcodes; debug stopcodes.
MPE Total memory parity errors.
RIB Total file system RIB errors.
POK This field shoew first the total number of POKE.'s performed and second
the job number of the last job to perform a POKE.
SCH Current SCHEDule value (see the DECsystem-10 Operating Systems Command
Manual under the SCHED command).
Next is a job summary:
Jobs Job totals in two fields: total jobs in use; total job capacity.
Log'd in Total number of jobs which are logged in.
Det Total number of jobs running detached (logged in or not).
Next, in as many columns as will fit (20 characters per column) is listed
the available disk file structures:
Struc The name of the disk file structure
Mnt The structure mount count
Free The total disk blocks free for that structure.
Finally the rest of the system display is devoted to I/O devices in use in
as many columns as will fit across the screen (at 13 characters per column).
Disks are not shown; magtape label dummy DDBs are not shown; and job command
terminals are not shown:
Dev The name of the device in use.
By The job number which owns the device.
How "A" if assigned to job; "I" if OPENed (or INITed) for I/O.
SYSDPY %434(562) Users Reference Manual Page 19
T display - network topology and message numbers
Following is a sample of the SYSDPY "T" display:
This is RW340A KL10 SYS#1279 01-Mar-79 04:13:41 UP:29:33:39
Node Neighbors OPR CTL LAR LAP LMS LMA LMR LMP
KL1279(10) 3, 2, 72 0 0 0 0 0 0
KL1026(26) 27, 15 15 15 15 15 15
CYNIC(66) 76, 3 3 3 3 3 3
KS4101(76) 66, 27, 11 11 15 15 245 245
CTCH22(22) 27, 3 3 3 3 3 3
NEXT(27) 26, 22, 2, 76, 3 3 3 3 4 4
RSX45(53) 2 2 2 2 7 7
DS401B(2) 10, 27, 101 101 101 101 211 211
DN87CP(3) 10, 53, 110 110 110 110 180 180
The first line is the standard SYSDPY "ID" line, containing the name of the
system (preceded by the header "This is"), the current date and time, and
finally the system uptime.
Following are the nodes currently up in the ANF network:
Node The name of the node, and the node number in parenthesis.
Neighbors The numbers of that node's immediate neighbors. If the node is
listed without any neighbors then it is a "sequential" node such as a
DN92 or an RSX-11M/DECnet node linked to the ANF network through a
DECnet compatible port.
OPR The TTY line number of the node's OPR terminal.
CTL The job number doing station ConTroL.
LAR Last ACK Received (NCL message number).
LAP Last ACK Processed (NCL message number).
LMS Last Message Sent (NCL message number).
LMA Last Message ACKed (NCL message number).
LMR Last Message Received (NCL message number).
LMP Last Message Processed (NCL message number).
SYSDPY %434(562) Users Reference Manual Page 20
\ display - network statistics
Following is a sample of the SYSDPY "\" display:
This is RW340A KL10 SYS#1279 01-Mar-79 04:13:53 UP:00:33:51
NTCOR= 3500 NTMAX= 4176 NTBAD= 5
Unnumbered CTL XMIT'ed RECV'ed Numbered CTL XMIT'ed RECV'ed
0 DAP/DATA 17637 14628 1 CONNECT 38 26
1 ACK 7707 5417 2 DISCONNECT 5 17
2 NAK 5 0 3 NEIGHBORS 8 8
3 REP 0 5 4 REQ CONFIG 8 7
4 START 6 3 5 CONFIG 7 14
5 STACK 3 5 6 DATA REQUEST 8 8425
6 NODE ID 2 2 7 STATION CTL 0 0
XMIT'ed=25360 Average=12.48/sec RECV'ed=20060 Average=9.87/sec
2**N 0% 20% 40% 60% 80% 99% 2**N 0% 20% 40% 60% 80% 99%
1 00% 1 00%
2 01% * 2 25% *******
3 34% ********* 3 68% *****************
4 05% ** 4 04% *
5 01% * 5 01% *
6 02% * 6 00%
7 25% ******* 7 00%
8 00% 8 00%
9 29% ******** 9 00%
The first line is the standard SYSDPY "ID" line, containing the name of the
system (preceded by the header "This is"), the current date and time, and
finally the system uptime.
The second line is general network values:
NTCOR Total monitor free core in use by NETSER.
NTMAX Maximum so far value of NTCOR.
NTBAD Number of bad network messages received.
The next portion of the display is devoted to the total number of network
messages received and transmitted, broken down by message type. Note that the
"Numbered CTL" messages are also counted in the "Unnumbered CTL" DAP/DATA
messages (they are the DAP messages). The total number of data messages can be
found by subtracting the total Numbered CTL messages from the DAP/DATA
Unnumbered CTL messages.
The next line summarizes total messages received and transmitted.
The last portion of the screen is a histogram of the data messages received
and transmitted, broken down as a function (log base 2) of the data message
size.
Note - This display does not scroll, and in particular, on terminals with
less than 20 (decimal) lines on the screen (e.g., VT50's) the histogram is not
displayed.