SIR Command
John_T_H...@msiinet.com
----------------------------------------------------------------------
INTERNAL ATTENTION ROUTINE COMMANDS - SIR
SIR, the abbreviation for System Information Report, is a VSE/ESA
Attention command that has mainly been developed for internal use, but
has been found very useful to system operators also. Any concerns,
recommendations and remarks and even complaints should be filed and
sent
via Internet to: "AX...@VNET.IBM.COM"
or to "DEIBMB5H at IBMMAIL"
It should, however, be clearly understood, that this command is NOT a
COMMITTED interface of any kind and that it is subject to change as
need arises. Differences that you may encounter on your system,
compared to the description given below, would be the result of such
changes. A sample is the VSE CPUID (ref: 1.0), which has been added
most recently and which might be missing in your version.
This description has been organized such, that the message prefix that
does show up on the operator console has been replaced by a reference
number (enclosed in parenthesis) that should be used in guiding you to
the detail description. Please don't argue about the values that you
will find in the samples, they may be constructed ones, just for the
purpose of this description, to make it easier to understand.
SIR ?
SIR HELP This command will provide HELP information about all the SIR
sub-commands which are currently available in your system.
?
HELP is the keyword requesting HELP for the SIR command(s). Its
main purpose is to show all the sub-commands and their syntax.
The command will produce similar output as shown in the sample
below.
sir ?
AR 0015 SIR COMMAND HELP
AR 0015 SIR DISPLAY SYSTEM INFORMATION
AR 0015 SIR SMF(=ON|OFF) SUBSYSTEM MEASUREMENT DATA
AR 0015 SIR MON(=ON|OFF(,FAST|BOUND) TD MONITORING DATA
AR 0015 SIR MIH(=nnnnnn|ON|OFF) DISPLAY/CHANGE MIH SETTING
AR 0015 SIR CHPID(=chpid) DISPLAY CHPID INFORMATION
AR 0015 SIR VENDOR DISPLAY VENDOR PRODUCT INFO
AR 0015 1I40I READY
EXPLANATION:
This command does provide information about the SIR sub-commands
which are currently available on your system. This command has
been provided for user convenience and it gives you the keywords
and options of any sub-command in a condensed form. For a detailed
description of the sub-commands you would have to refer to the
appropriate description later in this flash.
SIR (WITHOUT ANY OPERAND) may give you the information as outlined
below.
This command does provide general system information that might be
interesting for System-Administrator and -Operators as well. All
the timings are normally in the form
hhhh:mm:ss.nnn .
(hhhh=hours,mm=minutes,ss=seconds and nnn=milliseconds).
The command will produce similar output as shown in the sample
below.
sir
(1.0) CPUID VM = 3754320196720000 VSE = FFAF5AD596720000
(2.0) VM-SYSTEM = VM/ESA (LPAR) VER. 1 REL. 2.1 SERV.LVL. = 24C2
(3.0) PROCESSOR = 9672-37 USERID = VSEESA
(4.0) PROC-MODE = ESA IPL(140) 12/24/1995 19:05:36
CET
(5.0) AF-SYSTEM = VSE/AF VER. 6 REL. 1.0 SERV.LVL. =
DY43824
(6.0) IPL-PROC = $IPLESA JCL-PROC = $$JCL
(7.0) SUPVR = $$A$SUPX TURBO-DISPATCHER ACTIVE
(8.0) VIRTCPU = 0000:08:31.142 CP = 0000:04:05.431
(9.0) CPU-ADDR. = 0000(IPL) ACTIVE
(10.0) ACTIVE = 0000:00:13.602 WAIT = 0000:03:04.103
(11.0) PARALLEL= 0000:00:03.128 SPIN = 0000:00:00.002
(9.0) CPU-ADDR. = 0001 ACTIVE
(10.0) ACTIVE = 0000:00:12.514 WAIT = 0000:03:05.645
(11.0) PARALLEL= 0000:00:04.039 SPIN = 0000:00:00.003
(9.0) CPU-ADDR. = 0002 ACTIVE
(10.0) ACTIVE = 0000:00:02.783 WAIT = 0000:03:21.542
(11.0) PARALLEL= 0000:00:00.347 SPIN = 0000:00:00.000
(12.0) CPU timings MEASUREMENT INTERVAL 0000:03:28.494
(13.0) TASKS ATT.= 00000 HIGH-MARK = 00000 MAX = 00194
(14.0) DYN.PARTS = 00000 HIGH-MARK = 00000 MAX = 00018
(15.0)
(16.0) COPY-BLKS = 00258 HIGH-MARK = 00381 MAX = 01500
(17.0) CHANQ USED= 00004 HIGH-MARK = 00011 MAX = 00062
(18.0) PGIN TOT.= 0000013796 EXP.AVRGE.= 0000000044/SEC
(19.0) PGOUT TOT.= 0000000337
(20.0) UNC.= 0000000225 EXP.AVRGE.= 0000000018/SEC
(21.0) PRE = 0000000112 EXP.AVRGE.= 0000000002/SEC
(22.0) LOCKS EXT.= 0000000450 LOCKS INT.= 0000003636
(23.0) FAIL = 0000000006 FAIL = 0000000185
(24.0) LOCK I/O = 0000000000 LOCK WRITE= 0000000000
AR 0015 1I40I READY
EXPLANATION:
(1.0) CPUID VM = 3754320196720000 VSE = FFAF5AD596720000
This line contains the CPUID composed of the CPU Identification
Number with the Version Code followed by the Model Number that
your native system is running on. If you are running under VM, it
also gives you the VIRTUAL CPUID (field VSE) that has been defined
via the SET CPUID command under VM (appended with the real model
number). In the above example the native processor has a serial
number of '543201' and is a 9672 processor with version code '37'.
The VM command to set the CPUID was: SET CPUID AF5AD5.
(2.0) VM-SYSTEM = VM/ESA (LPAR) VER. 1 REL. 2.1 SERV.LVL. = 24C2
This line is OPTIONAL and only present if you are running under VM
and it gives information which version of VM you are running and
in addition will indicate if your VM system is running in an LPAR,
as is the case in this sample
(3.0) PROCESSOR = 9672-37 USERID = VSEESA
"PROCESSOR" specifies the Processor-Model-Number and the
Version-Code of the processor that you are running on.
"USERID" is OPTIONAL and only present under VM and is the
VM-userid of the VSE guest system that you are running
underneath VM.
(4.0) PROC-MODE = ESA IPL(140) 12/24/1995 19:05:36
CET
"PROC-MODE" specifies the processors mode of operation (the
different modes could be /370, XA, and ESA). ESA is the only
choice for VSE/ESA 2.1.x and higher.
"IPL" specifies the cuu (e.g 140) that has been used to get your
VSE/ESA system started. Appended is the date and time and
OPTIONALLY the time zone of when the VSE/ESA system was
IPL'ed. In this sample VSE/ESA was IPL'ed on the 24th of
December 1995 (Christmas surprise) at 5 minutes past 7 pm
Central European Time.
(5.0) AF-SYSTEM = VSE/AF VER. 6 REL. 1.1 SERV.LVL. =
DY43824
This line gives you detailed information about the AF-System
(Advanced Function System) that you are currently running. In the
sample we are running the AF Version 6 Release 1 Modification
Level 1 and the latest Supervisor-APAR that had been applied to
this system was DY43824. This information is extremely useful
when you have to place a problem call (PMR), which we hope will
never happen to you, in order to enable 1st- and 2nd-Level support
groups to check IBM's data base for known problems.
(6.0) IPL-PROC = $IPLESA JCL-PROC = $$JCL
"IPL-PROC" specifies the IPL-procedure that had been chosen by
either the system (based on specifications in the $ASIPROC or
based on defaults) or the operator. The named procedure was
executed to get your system started.
"JCL-PROC" specifies the JCL-procedure that had been chosen by
either the system (based on specifications in the $ASIPROC or
based on defaults) or the operator. The named procedure was
executed to get your partitions started.
(7.0) SUPVR = $$A$SUPX TURBO-DISPATCHER ACTIVE
"SUPVR" names the supervisor that had been loaded to control your
VSE/ESA system. Appended to the supervisor name you may find
information about the dispatcher that had been loaded. In this
sample the TURBO DISPATCHER, only available with VSE/ESA
2.1.x, had been loaded as opposed to the STANDARD DISPATCHER
that would have been loaded otherwise.
(8.0) VIRTCPU = 0000:08:31.142 CP = 0000:04:05.431
This line is OPTIONAL and only present when running under VM and
it gives you information about the real processor time that your
virtual system has consumed so far.
"VIRTCPU" is the total virtual time consumed by the virtual guest
system and
"CP" is that portion of the total virtual time (VIRTCPU) that CP
had spent specifically on behalf of the virtual CPU. In our
sample the total time consumed was 8 minutes, 31 seconds and
142 milliseconds whereof 4 minutes and 5.431 seconds where
consumed by CP itself.
(9.0) CPU-ADDR. = 0000(IPL) ACTIVE
This line is OPTIONAL and only present if you are running the
TURBO DISPATCHER and shows the different CPU addresses. This and
the following lines (ref:10.0 and 11.0) will be repeated for each
processor that does exist in your system. The information
subsequent to these line(s) (ref: 10.0 and 11.0) always applies to
the CPU address (CPU-ADDR) named last.
"CPU-ADDR" is the CPU address that the subsequent information (ref
10.0 and ref 11.0) is associated to. Appended to the CPU
address you find information about the current state of this
processor. In this example the processor is ACTIVE. Other
states could be INACTIVE, STOPPED, IN ERROR etc. The
information (IPL) is appended to one and only one processor
out of a range of processors that constitute an MP system and
it indicates that processor, that had been IPL'ed and it also
indicates that this is the processor that can't be STOPPED by
the operator.
(10.0) ACTIVE = 0000:00:13.602 WAIT = 0000:03:04.103
This line is OPTIONAL and only present if you are running the
TURBO DISPATCHER and it is related to the processor address
immediately preceding this line and it gives you information about
the processors CPU time consumption. In our sample it indicates
that the processor with the CPU address 0000 has ACTIVEly consumed
13 seconds and 602 milli- seconds and it had been in a WAITing
state for 3 minutes 4 seconds and 103 milli-seconds. The active
time does INCLUDE the SPIN time (ref:11.0).
(11.0) PARALLEL= 0000:00:03.128 SPIN = 0000:00:00.002
This line is OPTIONAL and only present if you are running the
TURBO DISPATCHER and it is related to the last preceding processor
address and gives information about the work-mix that this
processor was executing.
"PARALLEL" is the time that this processor executed work where
other processors could work concurrently. The higher this
value, the better the utilization of Multiple Processors (MP).
In our sample it indicates that the work-mix on this processor
was such that about 23% (3.128 seconds out of 13.602 seconds
ACTIVE time (ref 10.0) could be executed in parallel.
"SPIN" is the time that this processor spent in a very tight loop
waiting for system resources occupied by other processors
within the MP-system to be freed. In our sample processor
0000 spent 2 milliseconds in this tight loop.
(12.0) CPU timings MEASUREMENT INTERVAL 0000:03:28.494
This line is OPTIONAL and only present if you are running the
TURBO DISPATCHER and it defines the time that has elapsed since
the system was either IPLed, or since the measurement values have
been reset (SYSDEF TD,.....). All the values from the preceding
lines (ref:10.0 and 11.0) are related to this measuring interval.
In our sample the processor activity (for all the processors
mentioned before) was measured for 3 minutes and 28.494 seconds.
(13.0) TASKS ATT.= 00003 HIGH-MARK = 00008 MAX = 00194
This line informs you about the number of tasks that are, have
maximal been, and could ever be ATTACHED concurrently.
"TASK ATT." is the number of tasks currently ATTACHED.
"HIGH-MARK" is the maximum number of task that had ever been
attached concurrently since IPL.
"MAX" is the maximum number of task that can ever be attached
concurrently. It is the smaller value of: 256-32-NPARTS and
208 In the above example we have currently 3 tasks ATTACHED;
we had, at a certain point in time, a maximum of 8 task(s)
ATTACHED and we could attach a maximum of up to 194 tasks.
(14.0) DYN.PARTS = 00000 HIGH-MARK = 00000 MAX = 00018
This line informs you about the number of DYNAMIC PARTITIONS that
are, have maximal been, or could ever be ALLOCATED concurrently.
"DYN.PARTS" is the number of DYNAMIC PARTITIONS currently active.
"HIGH-MARK" is the maximum number of DYNAMIC PARTITIONS that had
been allocated concurrently since IPL.
"MAX" is the maximum number of DYNAMIC PARTITIONS that can be
allocated concurrently.
(15.0)
This line is left blank intentionally
(16.0) COPY-BLKS = 00258 HIGH-MARK = 00381 MAX = 01500
"COPY-BLKS" is the number of COPY-BLOCKS (required to properly
perform CCW-Translation) that currently are, have ever been
(since last IPL), or can maximal ever be used concurrently by
the currently running system.
In our sample we have currently 258 copy-blocks in use and
this value includes the number of copy-blocks that are being
held in use for FAST-CCW Translation.
"HIGH-MARK" gives the maximum number of copy-blocks that had ever
been used concurrently (since the last IPL) and it does
include the number of copy-blocks that had been held in use
for FAST- CCW Translation.
In our sample the maximum number of copy-block that had been
concurrently in use at a certain point in time (since last
IPL) was 381.
"MAX" specifies the number of copy-blocks that the currently
running system has allocated. This value can never be exceeded
without IPL'ing the system.
NOTE: If HIGH-MARK ever gets close to the MAX-value, this could
be an indication that you need more BUFSIZE entries to prevent I/O
performance degradation. You should set FASTTR=NO (Standard JCL
Option) and then watch these values again. If HIGH-MARK still
remains close to MAX-value, then you would probably have to
increase your BUFSIZE value.
If the HIGH-MARK never gets even close to the MAX-value, then it
would be an indication to run with a smaller number of BUFSIZE
entries (to save 72-bytes per entry) when you perform a new IPL.
(17.0) CHANQ USED= 00004 HIGH-MARK = 00011 MAX = 00062
"CHANQ USED" is the number of CHANnel Queue entries (one entry is
required per I/O operation) that currently are, have ever been
(since last IPL), or can maximal ever be used concurrently by
the currently running system. In our sample we have currently
4 channel-queue entries in use. Having a channel queue entry
in use does not necessarily indicate that an I/O operation is
currently ongoing.
"HIGH-MARK" gives the maximum number of channel queue entries that
had ever been used concurrently (since the last IPL). In our
sample the maximum number of copy-block that had been
concurrently in use at a certain point in time (since last
IPL) was 11.
"MAX" specifies the number of channel queue entries that the
currently running system has allocated. This value can never
be exceeded without IPL'ing the system.
NOTE: If HIGH-MARK ever gets close to the MAX value, this could
be an indication that you need more CHANQ entries to prevent I/O
performance degradation, and if it never gets even close to the
MAX value, then it would be an indication to run with a smaller
number of CHANQ entries (to save 32-bytes per entry) when you
perform a new IPL.
The next lines (ref: 18.0 through 21.0) are OPTIONAL and only
present if you are running a system with a PDS (Page Data Set)
having been allocated during IPL.
(18.0) PGIN TOT.= 0000013796 EXP.AVRGE.= 0000000044/SEC
This line informs you about the PAGE-IN activity since you have
IPL'ed the system. It is an indication of the total PAGE I/O
activity that was (PGIN TOT) or is (EXP.AVRGE.) ongoing.
"PGIN TOT." is the total number of PAGE-IN requests that have been
performed since the last IPL.
"EXP.AVRGE" is the EXPonential AVeRaGE PAGE-IN rate measured in
requests per second. Exponential Average means that the LATEST
PAGE-IN activity, measured during the latest measuring
interval (which is the time consumption between a predefined
number of PAGE-IN requests, which is processor dependant) is
taken into account with an equal weight, compared to the OLD
exponential average, in calculating the NEW exponential
average.
The formula is: ((LATEST-PAGE-IN/second)+(OLD-EXP.AVRGE))/2
to illustrate this, assume an: EXP.AVRGE.= 000000038/SEC
and assume a time difference from 200 milli-seconds between
a 1st and the 10th PAGE-IN request (assuming a processor
constant of 10), then the NEW EXP.AVRGE would calculate to:
LATEST PAGE-IN 200ms/10pgin = 50pgin/second
NEW EXP.AVRGE (50pgin/sec + 38pgin/sec) / 2 = 44
and thus our line would say: EXP.AVRGE.= 44
(19.0) PGOUT TOT.= 0000000337
This line informs you about the PAGE-OUT activity since you have
IPL'ed the system.
"PGOUT TOT." is the total number of PAGE-OUT requests that have
been performed since the last IPL. This line is the
accumulated value of the next two lines
(ref:20.0 and 21.0).
(20.0) UNC.= 0000000225 EXP.AVRGE.= 0000000018/SEC
"UNC." gives the number of UNCONDITIONAL PAGE-OUT requests which
have either been explicitly requested, or which have been
initiated by the PAGE MANAGER to free one or more PAGE FRAMEs.
"EXP.AVRGE" is the EXPonential AVeRaGE UNC.PGOUT rate measured in
requests per second. Exponential Average means that the LATEST
UNC.PGOUT activity, measured during the latest measuring
interval (which is the time consumption between a predefined
number of PAGE-IN requests, which is processor dependant) is
taken into account with an equal weight, compared to the OLD
exponential average, in calculating the NEW exponential
average.
The formula is: (LATEST-UNC.PGOUT/second)+(OLD-EXP.AVRGE)/2
A detailed sample that you may want to refer to is given above
(ref:18.0).
(21.0) PRE = 0000000112 EXP.AVRGE.= 0000000002/SEC
"PRE" gives the number of PRE PAGE-OUT requests which have been
initiated by the PAGE MANAGER to ensure PAGE FRAME
availability when a PAGE-IN request should be received. These
values might serve as an indication of how often pages are
being referenced by the system. Pages not being referenced any
more, will be subject to destaging (PRE-PAGE-OUT) if need
arises.
"EXP.AVRGE" is the EXPonential AVeRaGE PRE.PGOUT rate measured in
requests per second. Exponential Average means that the LATEST
PRE.PGOUT activity, measured during the latest measuring
interval (which is the time consumption between a predefined
number of PAGE-IN requests, which is processor dependant) is
taken into account with an equal weight, compared to the OLD
exponential average, in calculating the NEW exponential
average.
The formula is: (LATEST-PRE.PGOUT/second)+(OLD-EXP.AVRGE)/2
A detailed sample that you may want to refer to is given above
(ref:18.0).
The next lines (ref: 22.0 through 24.0) give you some information
about the LOCKTAB and/or LOCKFILE (DLF) usage and it tells you
about conflicts that have been encountered due to attempts to
access certain resources concurrently.
(22.0) LOCKS EXT.= 0000000450 LOCKS INT.= 0000003636
This line contains the total number of LOCK (USE) requests that
have been issued since the last IPL.
"LOCKS EXT." is the total number of EXTERNAL locks that had been
requested. EXTERNAL means that the requestor has requested a
resource in a SHARED environment for controlled access. It
DOES NOT indicate that you are RUNNING a shared environment.
In the above sample there has been a total of 450 LOCK request
to guarantee controlled access by other processors that do
have access to the specified resource.
"LOCKS INT." is the total number of INTERNAL locks that had been
requested. INTERNAL means that the requestor has requested a
resource in his own system to be locked for controlled access.
Internal LOCKS are those for which controlled access by other
tasks within this system is granted. In our sample we had
3636 LOCK (USE) requests with the attribute INTERNAL (grant
controlled access by other tasks).
(23.0) FAIL = 0000000006 FAIL = 0000000185
This line contains the total number of UNSUCCESSFUL LOCK attempts
either EXTERNAL (left column) or INTERNAL (right column)
"FAIL" is the total number of LOCK requests that had been
unsuccessful due to access constrains with other requestors.
In the above example, 6 EXTERNAL LOCK requests (out of 450
(ref:22.0)) had been unsuccessful and as such had to be
re-issued. Similar, 185 out of 3636 INTERNAL LOCK requests
had been unsuccessful and would have had to be re-issued.
(24.0) LOCK I/O = 0000000000 LOCK WRITE= 0000000000
This last line gives information about the LOCKFILE I/O accesses.
"LOCK I/O" is the total number of I/O operations, whereby the LOCK
request itself might have been success-or unsuccessful, that
had been issued to the LOCKFILE.
"LOCK WRITE" is the total number of I/O operations where an
external lock was either obtained, or freed. The greater the
difference between these two values is, the more unbalanced is
the workload among the different processors that constitute
the SHARED environment. The fact that in our sample the LOCK
I/O is 00000000, indicates that we where running a NON-SHARED
system.
SIR SMF(=ON|OFF) This command allows to retrieve data from the
Subsystem Measurement Facility. It is mainly intended to assist in
finding I/O performance problems. Since the Channel Subsystem does
maintain these counters without posting an overflow condition, it
is recommended to not run this activity report for too long.
Especially the I/O-CNT should NOT EXCEED 65,536 to ensure reliable
values.
SMF is the keyword indicating that Subsystem Measurement Facility
data is requested and, if specified without any further
options will retrieve that data, assuming that SMF had been
activated before.
SMF=ON indicates that you want to activate the Channel Subsystem
Measurement Facility which is a prerequisite to retrieve SMF
data.
SMF=OFF indicates that you want to deactivate the Channel
Subsystem Measurement Facility.
The following sample should illustrate the functional
capabilities.
The command will produce similar output as shown in the sample
below.
sir smf
(1.0) SUBSYSTEM MEASUREMENT FACILITY IS INACTIVE
sir smf=on
(2.0) SUBSYSTEM MEASUREMENT FACILITY HAS BEEN ACTIVATED
(3.0) NO SUBSYSTEM MEASUREMENT DATA YET AVAILABLE
-------here some I/O activity was forced-----------
AR 0015 2FE 90E5 FBAV UNUSED DOWN
AR 0015 2FF 90E5 FBAV UNUSED DOWN
AR 0015 330 6E 3390-06 BAM330 CMS-D DOWN
AR 0015 480 54C3 TAPE UNUSED NOT OPER.
AR 0015 481 54C3 TAPE UNUSED NOT OPER.
AR 0015 490 52D3 TAPE UNUSED NOT OPER.
AR 0015 491 52D3 TAPE UNUSED NOT OPER.
AR 0015 1I40I READY
sir smf
(4.0) DEVICE I/O-CNT QUEUED CONNECT DISCONN TOTAL
(4.1) msec/SSCH msec/SSCH msec/SSCH
msec/SSCH
(5.0)
(6.0) 100 5 0.537 3.507 3.993
8.038
(6.0) 101 5 0.204 10.214 1.459
11.878
(6.1) 190 5 0.153 2.585 0.000
2.739
(6.0) 191 5 0.230 9.088 3.763
13.081
(6.0) 19B 5 0.179 7.987 0.870
9.036
(6.1) 19D 5 0.153 2.611 0.000
2.764
(6.0) 19E 5 0.230 6.579 1.331
8.140
(6.2) 110 13 3.810 2.491 17.723
24.024
(6.0) 130 3 0.256 2.816 0.000
3.072
(6.0) 131 3 0.213 3.456 3.029
6.698
(6.0) 132 3 0.170 5.290 2.602
8.064
(6.0) 133 3 0.170 4.138 4.608
8.917
(6.0) 140 3 0.170 1.706 3.242
5.120
(6.0) 230 3 0.213 5.546 2.688
8.448
sir smf=off
AR 0015 1I40I READY
sir smf
(1.0) SUBSYSTEM MEASUREMENT FACILITY IS INACTIVE
EXPLANATION:
(1.0) SUBSYSTEM MEASUREMENT FACILITY IS INACTIVE
This message is telling us that the Subsystem Measurement Facility
(SMF) is inactive, or has been deactivated (SIR SMF=OFF).
The operator would have to issue the SIR SMF=ON command to get the
facility initialized properly.
(2.0) SUBSYSTEM MEASUREMENT FACILITY HAS BEEN ACTIVATED
This is the acknowledgement message that the Measurement Facility
has now been successfully initialized.
(3.0) NO SUBSYSTEM MEASUREMENT DATA YET AVAILABLE
This message indicates that the Subsystem Measurement Facility is
active, but that NO measurement data is yet available. Either no
I/O operation has been issued since the Measurement Facility has
been activated, or the device is not being supported by the
measurement facility.
(4.0) DEVICE I/O-CNT QUEUED CONNECT DISCONN TOTAL
(4.1) msec/SSCH msec/SSCH msec/SSCH
msec/SSCH
This line is a heading line describing the contents of the
vertical fields below the headings subject. The heading line as
well as the sub- heading line will be repeated after 16 detail
lines to make the console output easier to read. Timing values are
given in milli-seconds per SSCH (START SUBCHANNEL).
"DEVICE" specifies the cuu address of the device that is being
measured.
"I/O-CNT" specifies the number of I/O operations that had been
used in calculating the measurement data in the subsequent
fields. Watch-out, the I/O-CNT wraps at 65,536 back to 0000
but the timings will be accumulated for about 153 hours which
will cause inconsistant data once the I/O-CNT wraps. To
prevent such inconsistent data, it is recommended to set SIR
SMF=OFF once the I/O-CNT get close to that wrap value. A
subsequent activation (SET SMF=ON) will start with all fields
reset to zero.
"QUEUED" this field contains the average time (based on the
I/O-CNT) that an I/O request was queued in the Channel
Subsystem. An I/O request would be held pending (queued) in
the Channel Subsystem if e.g. a channel or a Control Unit (CU)
is currently busy. The values in this column would thus
indicate channel or CU contention.
"CONNECT" this field contains the average time (based on the
I/O-CNT) that a device is logically connected to a channel for
purposes of transferring information between it and the
Channel Subsystem.
"DISCONN" this field contains the average time (based on the
I/O-CNT) that a device is logically disconnected from the
Channel Subsystem while the device is still busy and has not
yet presented primary interrupt (Channel End) status. This
value does NOT include the time between Channel End (CE) and
Device End (DE).
"TOTAL" this field contains the average time (based on the
I/O-CNT) of a complete I/O operation and is actually the sum
of QUEUED, CONNECT and DISCONN. It should be used to quickly
check if there are any I/O-device problems.
(6.0) 100 5 0.537 3.507 3.993
8.038
This is a sample line telling us that the average time, that had
been measured based on 5 SSCH-instructions, issued to the device
with the cuu=100, had been 8 milli-seconds and 38 micro-seconds in
total. From this 8.038 milli-seconds, the request was QUEUED in
the Channel Subsystem for about half a milli-second (0.537), was
doing data transfer operations for 3.507 milli-seconds and was
executing SEEK/SEARCH type operations (DISCONNected) for about
3.993 milli-seconds. This sample line was for a DASD device and
might be considered usual DASD timings. The other lines (ref:
6.0) are similar.
(6.1) 190 5 0.153 2.585 0.000
2.739
This is a sample line telling us that the average time, that had
been measured based on 5 SSCH-instructions, issued to the device
with the cuu=190, had been 2 milli-seconds and 739 micro-seconds
in total. From this 2.738 milli-seconds, the request was QUEUED
in the Channel Subsystem for about 153 micro-seconds, was doing
data transfer operations for 2.585 milli-seconds and was NOT
executing SEEK/SEARCH type operations (DISCONNected). This sample
might be considered a typical sample for CACHED DASD I/O
operations with CACHE HITS. The other lines (ref: 6.1) are
similar.
(6.2) 110 13 3.810 2.491 17.723
24.024
This is a sample that should be examined more carefully. It is
telling us, that, based on 13 I/O-instructions the average
execution time was about 24 milli-seconds (TOTAL). The time it
took the Channel Subsystem to actually get the device started was
about 3.810 milli-seconds. This, compared to the figures before,
is quite some time, that needs interpretation. In the given
scenario, where the 110 was a MINI-DASD (under VM) it could mean
that there was some device contention on the REAL device. The long
disconnect (DISCONN) time could be an indication that there might
be lost revolutions or RPS (Rotational Position Sensing) misses on
this device, or, it could indicate that there is heavy SEEK
activity ongoing. I would suggest that you watch such devices for
more than just a few I/O instructions and then draw your
conclusion(s).
If the TOTAL time ever gets higher than about 30 milli-seconds,
then this should raise your alert flag.
SIR MON(=ON|OFF(,FAST(,BOUND))) This command does provide monitoring
data and is available with the TURBO DISPATCHER only.
MON is the keyword required to retrieve monitoring data.
"ON" indicates that monitoring is to be started. If specified
without any further operands, it will cause the usage of
Supervisor Calls (SVC) only to be monitored.
"FAST" indicates, that, in addition to the supervisor calls,
also the FAST-SVC calls (SVC-6B) be explicitely monitored
by their function code.
"BOUND" indicates, that, in addition to the supervisor calls,
also the WAIT conditions (Bound states) be explicitely
monitored.
"OFF" indicates that monitoring is to be stopped. If
specified without any further operands, it will cause all
monitoring to be stopped. If FAST or BOUND have been
specified, monitoring will be deactivated partially.
The command will produce similar output as shown in the sample
below.
SIR MON
AR 0015 MONITORING REPORT (SVC)
AR 0015 SVC-00 = 390 SVC-04 = 144 SVC-07 =
384
AR 0015 SVC-0D = 201 SVC-0F = 313 SVC-16 =
44
AR 0015 SVC-1D = 57 SVC-21 = 292 SVC-22 =
59
AR 0015 SVC-39 = 2 SVC-3D = 400 SVC-3E =
413
AR 0015 SVC-4B = 55 SVC-53 = 4 SVC-54 =
8
AR 0015 SVC-63 = 27 SVC-68 = 1 SVC-6B =
770
AR 0015 SVC-6E = 546 SVC-70 = 29 SVC-72 =
1
AR 0015 SVC-7C = 2 SVC-83 = 57
AR 0015 MONITORING REPORT (FAST-SVC)
AR 0015 FC-02 = 121 FC-06 = 49 FC-1B =
48
AR 0015 FC-1E = 63 FC-31 = 1 FC-3F =
55
AR 0015 FC-40 = 41 FC-42 = 5 FC-44 =
9
AR 0015 FC-46 = 55 FC-47 = 56 FC-49 =
5
AR 0015 FC-4A = 8 FC-4B = 8 FC-4E =
17
AR 0015 FC-4F = 10 FC-51 = 35 FC-59 =
2
AR 0015 FC-5D = 1 FC-5F = 1 FC-62 =
4
AR 0015 FC-67 = 8 FC-73 = 60 FC-74 =
6
AR 0015 FC-7D = 2 FC-7E = 1 FC-85 =
12
AR 0015 FC-90 = 98 FC-96 = 1 FC-AA =
1
AR 0015 MONITORING REPORT (BOUND STATE)
AR 0015 BND-49 = 3 BND-50 = 12429 BND-80 =
220
AR 0015 BND-82 = 440 BND-85 = 179 BND-87 =
1
AR 0015 1I40I READY
SIR MIH(=NNNNNN|ON|OFF) This command does provide facilities to
manipulate the invocation of the MIH (Missing Interrupt Handler)
process in VSE, or to show the current settings of the MIH cycle
time.
The MIH process in VM, if running under VM, remains UNAFFECTED by
this command.
MIH is the keyword indicating that MIH constants are to be
displayed/altered.
"NNNNNN" is the numeric value in seconds, after which the MIH
process is to be (re-)initiated. This is called the MIH
cycle time which is normally set to 180 seconds. Any value
between 1 and 999999 will be accepted.
"ON" indicates that the MIH process with the existing
constants is to be re-activated.
"OFF" indicates that the MIH process is to be totally
bypassed, and as a result no more lost interrupts, if any,
will be recognized.
The following sample should illustrate the functional
capabilities.
The command will produce similar output as shown in the sample
below.
sir mih
AR 0015 MIH ON INTERVAL= 5 SECONDS
sir mih=off
AR 0015 MIH PROCESSING NOW INACTIVE
sir mih
AR 0015 MIH OFF INTERVAL= 5 SECONDS
sir mih=180
AR 0015 MIH TIME INTERVAL SET TO 180 SECONDS
sir mih
AR 0015 MIH OFF INTERVAL= 180 SECONDS
sir mih,on
AR 0015 MIH PROCESSING NOW ACTIVE
sir mih
AR 0015 MIH ON INTERVAL= 180 SECONDS
SIR CHPID This command does provide information about the Channel
Pathes (CHPID) attaching to your system. This command is available
on native systems only and will be considered INVALID if issued
under VM.
CHPID is the keyword required to retrieve Channel path
information.
The command will produce similar output as shown in the sample
below.
sir chpid
AR 0015 CHPID CHLA SWLA LSN CHANNEL-PATH-DESCRIPTION
AR 0015 00 PARALLEL BYTE-MULTIPLEXER
AR 0015 01 PARALLEL BLOCK-MULTIPLEXER
AR 0015 02 PARALLEL BLOCK-MULTIPLEXER
AR 0015 03 PARALLEL BLOCK-MULTIPLEXER
AR 0015 04 PARALLEL BLOCK-MULTIPLEXER
AR 0015 05 PARALLEL BLOCK-MULTIPLEXER
AR 0015 06 PARALLEL BLOCK-MULTIPLEXER
AR 0015 07 PARALLEL BLOCK-MULTIPLEXER
AR 0015 08 PARALLEL BLOCK-MULTIPLEXER
AR 0015 09 PARALLEL BLOCK-MULTIPLEXER
AR 0015 0A PARALLEL BLOCK-MULTIPLEXER
AR 0015 0B PARALLEL BLOCK-MULTIPLEXER
AR 0015 18 01 SERIAL POINT-TO-POINT
AR 0015 19 FIBER EXTENDED CHANNEL
AR 0015 18 01 SERIAL POINT-TO-POINT
AR 0015 19 FIBER EXTENDED CHANNEL
AR 0015 1A 01 FIBER EXTENDED CHANNEL
AR 0015 1B SERIAL CHANNEL PATH (INIT)
AR 0015 CHPID CHLA SWLA LSN CHANNEL-PATH-DESCRIPTION
AR 0015 1C C4 FE 01 SERIAL SWITCHED POINT-TO-POINT
AR 0015 1D C5 FE 01 SERIAL SWITCHED POINT-TO-POINT
AR 0015 1E C6 FE 01 SERIAL SWITCHED POINT-TO-POINT
AR 0015 1F 01 SERIAL POINT-TO-POINT
AR 0015 1I40I READY
EXPLANATION:
"CHPID" specifies the CHPID of the channel path to which the
information in this row applies.
"CHLA" CHannel Link Address indicates the link address assigned to
the channel-subsystem link-level facility of the specified
channel path (CHPID). This is used by control units as the
destination link address when attempting to communicate with
the channel subsystem on the specified channel path.
"SWLA" SWitch Link Address contains the link address assigned to
the control unit link-level facility of the dynamic-switch.
This is used as the destination link address when attempting
to communicate with the dynamic-switch control unit on the
specified channel path (CHPID).
"LSN" Logical Switch Number gives the logical switch number of the
switch, if any, that is attached through the corresponding
Channel-Path Link Address (CHLA).
John T. Hudson
S390 Systems Engineer
MSI Systems Integrators
Office: 612-317-4612
Fax: 612-317-4619
Pager 800-621-2962
jhu...@msiinet.com
Martin Truebner c/o Pi GmbH
>> Here is a good description of the SIR command and what it ..<<
thank you for posting the excellent description of the SIR-displays.
Two questions to ??(is the AXEL in the E-mail AXEL Piper?):
1.) What are the counters counting that are displayed if MON is on?
I am pretty sure they are state-changes, but what is what?
2.) How can I participate in the counting of SVC's?
two reasons for asking:
a.) I have my own that identifies submitters of jobs to the rdr (which
is simply not possible with the new technique of power reading
everything in and then passing the read data (cards) to the
PWR-reader-exit
b.) I would like to see the number of SVC's issued to contact ADABAS.
Loud thinking: If me (doing it just for the sake of correct stats)
can get stats about calls to "my" SVC--
Users of SAG-ADABAS might get their act together and ask SAG
to do the same.
Martin userid: mar...@pi-sysprog.de
homepage at HTTP://www.pi-sysprog.de
PUN2IPT to control compiles from a REXX-proc (and gain performance)
Fran Hensler
>Here is a good description of the SIR command and what it displays.....
Thank you, John, for that excellent presentation. Especially useful was the
description of the PARALLEL and SPIN values.
/Fran Hensler at Slippery Rock University of Pennsylvania f...@sruvm.sru.edu