SOS Global Process Stops

The Global Process Stops screen provides information that helps to determine why processes are being impeded. This information shows why system, session and job processes are giving up use of CPU. Keep in mind that a process will continue to use the CPU until one or more events occur to impede its progress.

Figure 18.1 SOS Global Process Stop screen

To access the Global Process Stop screen from the Global Summary screen:

Type S from the SOS Enter command: prompt to view the Screen Selection Menu screen.

From the Screen Selection Menu screen, enter S (Global Stop Screen). The Global Process Stop screen will display (refer to Figure 18.1).

A simple example of this is when a single program reads principal balances from disc and calculates dividends. The CPU provides service to the program unless the program indicates that it no longer needs any or unless a stop event occurs. One of the most common stop events occurs when a trip to disc is necessary for the process to continue. The CPU recognizes that it is poor use of time to be monopolized by a single process (waiting for the disc I/O to occur) when it can be more productive (servicing other processes, housekeeping, etc.).

Although there are more stop reasons (fifty-five and an “other” category!) it is common to see a small number of reasons representing the majority of stop activity. Due to the scarcity of internal information available on MPE/iX we can only provide surface definitions for most of the stop reasons. If you notice a consistent amount of stop activity in a counter that you cannot determine or account for please call us. We will keep you updated on any new information concerning these obscure activities via bulletins and future releases.

Global Stop Screen Keys

All of the Global Stop Screen keys are identical to the keys of the CPU Detail Screen. See "CPU Detail Screen Keys" for details.

Global Process Stop Data Items

Notice that there are two statistics for each stop reason: the Current Stop percentage and the Average Stop percentage (in brackets). The average represents a value that has been accumulated since SOS/3000 was initiated or since the last time the RESET TOTALS function key was pressed, whichever occurred last.

Table 18.1 SOS Global Process Stop Data Items

Data Item	Description
NM Code page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a native mode code page fault.
NM stk page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a native mode stack page fault.
NM trns page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a native mode transient page fault (heap, swapable table).
File page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a file page fault.
CM code page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a compatibility mode code page fault.
CM stk page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a compatibility mode stack page fault.
CM trns page flt	The percentage of all stops due to processes blocking (stopped) for a disc I/O to complete on a compatibility mode transient page fault.
Terminal read	The percentage of all stops due to processes blocking (stopped) for terminal reads.
Terminal write	The percentage of all stops due to processes blocking (stopped) for terminal writes. These include events such as console messages or general TELL and WARN messages to other sessions.
Disc I/O	The percentage of all stops due to processes blocking (stopped) for I/O to disc devices.
Other I/O	The percentage of all stops due to processes blocking (stopped) for I/O to non-disc devices. These devices include tape drives and printers.
IOWAIT	The percentage of all stops due to processes blocking (stopped) on IPC (interprocess communication) with transaction completed as one option. An example of this is use of the IOWAIT intrinsic.
SIR	The percentage of all stops due to processes blocking (stopped) for SIR’s (System Internal Resource). It is used much like taking a number at a hardware store. SIR’s provide a way to control which processes get access to special system services and tables. The system Group/Account is an example of one such resource.
RIN	The percentage of all stops due to processes blocking (stopped) for RIN’s (Resource Identification Number). RIN’s are used to coordinate file locking for files that are accessed by multiple processes.
Mem Mgr prefetch	The percentage of all stops due to processes blocking (stopped) for a memory manager prefetch action to a disc device.
Quantum used	The percentage of all stops due to the process consuming its entire time quantum.
Uint16 timer	The percentage of all stops due to processes blocking (stopped) for a time-out or pause (e.g., PAUSE intrinsic) with one second or less.
Father	The percentage of all stops due to processes blocking (stopped) which are waiting to be wakened by one or more son processes.
Semaphore ctl blk	The percentage of all stops due to processes blocking (stopped) for a control block on a semaphore.
Son	The percentage of all stops due to processes blocking (stopped) which are waiting to be wakened by their father process.
Data comm	The percentage of all stops due to processes blocking (stopped) for data communication.
Operator reply	The percentage of all stops due to processes blocking (stopped) which are waiting for an Operator reply (RIT wait).
Disp preempt	The percentage of all stops due to preemption by dispatcher to work on higher priority system processes (power failure, grey page cleanup, replenish critical pool, fetch IO, or system fetch).
Port	The percentage of all stops due to processes blocking (stopped) for a port. This is the default IPC wait.
Mail	The percentage of all stops due to processes blocking (stopped) for MAIL. MAIL is an older type of an IPC (inter-process communication) interface that existed before the message file implementation.
Junk	The percentage of all stops due to processes blocking (stopped) for a JUNK wait. This is a special system process stop activity.
Message file	The percentage of all stops due to processes blocking (stopped) for a MESSAGE. A MESSAGE is the basic IPC message file wait reason.
Impede	The percentage of all stops due to processes blocking (stopped) for file impedes. An impede is used by the file system to synchronize access to files.
Break	The percentage of all stops due to processes blocking (stopped) because they are in break mode. This means that either the BREAK key was pressed at a terminal or the BREAKJOB command was issued against an executing job.
Wait queue	The percentage of all stops due to processes blocking (stopped) for PORTS. This happens when a system table management runs out of entries and waits for additional space to be allocated.
Mem Mgr wait	The percentage of all stops due to processes blocking (stopped) for proper disc I/O synchronization. This excludes activity such as a user I/O requesting a POST.
Port absent	The percentage of all stops due to processes blocking (stopped) for requested ports from the PORTS facility to become available.
File blocked	The percentage of all stops due to processes blocking (stopped) on a port when posting pages to disc through the call CM_POST.
File unblocked	The percentage of all stops due to a wait for a file to be unblocked.
Storage mgmt	The percentage of all stops due to processes blocking (stopped) on a port through the storage management facility.
User debug	The percentage of all stops due to processes (CI’s - Command Interpreters) blocking (stopped) due to breakpoint contention. This is usually because of using DEBUG breakpoints.
I/O config	The percentage of all stops due to processes blocking (stopped) devices are being configured or released.
PFP reply	The percentage of all stops due to processes blocking (stopped) because the port facility process needs to be created, initialized or checked.
DB monitor	The percentage of all stops due to processes blocking (stopped) because the database monitor (for SQL) is waiting for the DB_CLEAN_UP to finish cleaning up the aborted processes before closing the database.
Fill disc	The percentage of all stops due to processes blocking (stopped) HLIO is waiting for the master MIB because the new file extent in secondary storage needs to be initialized with fill characters for all virgin pages.
HLIO	The percentage of all stops due to processes blocking (stopped) because the HLIO is aborting the I/O.
TIO	The percentage of all stops due to processes blocking (stopped) because Terminal I/O (TIO) fast write in DTS is waiting for a reply message from the device manager or buffer management.
Mem Mgr post	The percentage of all stops due to processes blocking (stopped) which are waiting for I/O completion when explicitly posting pages to disc.
Signal timer	The percentage of all stops due to processes blocking (stopped) for a delay or timer on a standard signal port.
CPU preempt	The percentage of all stops due to preemption by a higher priority process due to process awakening (IPC wait other than disc I/O completion).
Disc I/O preempt	The percentage of all stops due preemption by a higher-priority process due to disc I/O completion. This includes page fault, post, and prefetch.
Priority preempt	The percentage of all stops due to preemption by a higher priority process due to priority boosting or dropping.
SQL lock	The percentage of all stops due to processes blocking (stopped) which are waiting to acquire an SQL lock. This lock is required for user data (a tuple, a page or a relation).
SQL latch lev 1	The percentage of all stops due to processes blocking (stopped) for a level 1 latch. A latch is used to coordinate access to its run-time data structures. Each latch has a level associated with it and is used for deadlock prevention.
SQL latch lev 2	The percentage of all stops due to processes blocking (stopped) for a level 2 latch. A latch is used to coordinate access to its run-time data structures. Each latch has a level associated with it and is used for deadlock prevention.
SQL latch lev 3	The percentage of all stops due to processes blocking (stopped) for a level 3 latch. A latch is used to coordinate access to its run-time data structures. Each latch has a level associated with it and is used for deadlock prevention.
SQL latch lev 4	The percentage of all stops due to processes blocking (stopped) for a level 4 latch. A latch is used to coordinate access to its run-time data structures. Each latch has a level associated with it and is used for deadlock prevention.
SQL buffer	The percentage of all stops due to processes blocking (stopped) to change the state of a page in the SQL buffer. The buffer is used to hold a page (4K) of user data. The page can be in a number of states (for example being updated in transit-in, transit-out-of). When a process requests that a page be placed in a state that conflicts with its current state, the process blocks.
Long timer	The percentage of all stops due to processes pausing for two or more seconds.
Mem Mgr freeze	The percentage of all stops due to processes blocking (stopped) on freeze and corner cases other than page fault prefetch and freeze. This counter predominantly blocks on freeze.
Other	This category includes stop events that are not covered under any of the above-mentioned reasons. HP has not documented what these reasons are. Our experience shows that small numbers in these “Other” events occur relative to the others described above. This counter will usually be zero.