1 03/26/85 monitor_cache
2
3 Syntax as a command: monitor_cache cpu_tag -control_args
4
5
6 Function: initiates and controls automatic monitoring of cache memory
7 error data saved during normal fault processing. This command
8 constantly monitors the cache memory error data to provide a warning
9 when error rates become excessive. To compare error rate threshold
10 values, use a per process threshold or the -priv control argument
11 described below. For information pertaining to changing the per-system
12 defaults and setting up per-process values that are different from the
13 per-system defaults, see "Changing the Threshold Values" below. To
14 display the current threshold values, use the display_cache_threshold
15 command.
16
17
18 Arguments:
19 cpu_tag
20 identifies the tags of the CPUs whose cache error rates should
21 be monitored. If no CPUs are identified, all possible CPUs will be
22 monitored.
23
24
25 Control arguments:
26 -brief, -bf
27 suppresses the "CPU below cache error threshold" message when the
28 error rate is within the specified threshold limits. This does not
29 suppress the warning when the error rate is above the threshold
30 values. This is the default.
31 -cpu cpu_list
32 an alternate method of specifying the list of the tags of the CPUs
33 whose cache error rates should be monitored.
34 -long, -lg
35 emits a "CPU below cache error threshold" message on the user_output
36 I/O switch.
37
38
39 -priv
40 causes warning messages of "Cache above error threshold for CPU" to
41 be written into the syserr_log and the bootload console with an
42 audible alarm. This control argument sends the polling message
43 described under -long above to the syserr_log instead of the
44 user_output I/O switch. You must have re access to the hphcs_ gate
45 to use this control argument.
46 -start, -sr
47 resumes monitoring for the CPUs specified in the cpu_tag argument or
48 the -cpu control argument. Monitoring will continue with the next
49 scheduled cycle.
50
51
52 -stop, -sp
53 stops monitoring for the CPUs specified in the cpu_tag argument or
54 the -cpu control argument. If no CPU tags have been specified, then
55 monitoring for all CPUs is terminated.
56 -time N, -tm N
57 specifies the monitoring interval in minutes, where N is a decimal
58 integer. The default time is 15 minutes.
59
60
61 Access required: You must have re access to the phcs_ gate to use this
62 command.
63
64
65 Notes on changing the threshold values: The per-system and
66 per-process default threshold values are defined in
67 >system_library_tools>cache_threshold_defaults. This segment is
68 created by a cds source segment of the same name. To change the system
69 default values, you must change this source segment, recompile it, and
70 install the generated object in >tools. A per-process threshold that
71 is different from the per-system values may be created by performing
72 the same operations, except that the per-process threshold should be
73 found in the process' object search rules before >tools or the
74 threshold may be specifically initiated.
75
76
77 Notes: The monitor interval is closely associated with the threshold
78 values in that the threshold is specified in terms of an acceptable
79 error rate N in X time. The default threshold values are expressed as
80 a per_hour_error rate.