The system activities package monitors the various counters in the kernel that measure system usage. You can then optimize the system based on the system activity package report. If you use the system activity package, see the SUPER-UX System Administrator's Guide.
The sar command of the system activity package reports on the following contents.
| Home |
|---|
sar -a reports the use of file access operations. The following UNIX operating system routines are reported:
An example of sar -a output, with a 30 second sampling interval, follows:
12:41:40 iget/s namei/s dirbk/s
12:42:10 4 1 3
12:42:40 2 1 1
12:43:10 5 2 3
Average 4 1 3
|
The larger the values reported, the more time the UNIX kernel is spending to access user files. This indicates how heavily programs and applications are using the file system(s). The -a option is helpful for understanding how disk-dependent the application system is; it is not used for any specific tuning step.
| Home |
|---|
sar -b reports the following buffer activity via network.
When the system load is high, %rcache may report a minus value because of re-reading the same block which has been read ahead.
An example of sar -b output follows:
16:32:57 bread/s lread/s %rcache bwrit/s lwrit/s %rcache pread/s pwrit/s
12:33:07 3 39 93 1 16 91 0 0
12:33:17 4 40 90 2 16 87 0 0
12:33:27 4 41 90 3 7 64 0 0
Average 4 40 91 2 13 84 0 0
16:32:57 aread/s awrit/s
12:33:07 1 0
12:33:17 3 2
12:33:27 2 1
Average 2 1 |
This example shows that the buffers are not causing any bottlenecks, because all data is within acceptable limits.
| Home |
|---|
sar -B reports the following buffer activity of SFS file system.
An example of sar -B output follows:
12:41:40 sbrd/s slrd/s strd/s %srche %srnob sbwt/s slwt/s stwt/s %swche %swnob
12:42:10 2 3 12 33 75 3 5 21 40 76
12:42:40 3 4 18 25 77 3 6 19 50 68
12:43:10 1 2 16 50 87 2 2 17 0 88
Average 2 3 15 33 80 3 4 19 25 79 |
The fraction of cache hit is not important on SFS file system. The unused SFS file system buffer reads or writes for better performance. So consider how to use the SFS file system if %srnob and %swnob values are low.
| Home |
|---|
Reads and writes account for about half of the total system calls, although this varies greatly with the activities being performed by the system.
An example of sar -c output follows:
18:33:04 scall/s sread/s swrit/s fork/s exec/s rchar/s wchar/s
18:33:35 38 16 6 0.03 0.03 6089 1638
18:34:05 39 16 4 0.07 0.07 6123 1602
18:34:35 38 17 5 0.17 0.17 6042 1704
Average 38 16 5 0.09 0.09 6085 1648 |
sar -m reports on interprocess communication activities. Message and semaphore calls are reported as follows:
An example of sar -m output follows:
15:16:58 msg/s sem/s
15:17:32 0.00 0.00
15:18:02 0.00 0.00
15:18:32 0.00 0.00
Average 0.00 0.00 |
These figures usually are zero (0.00) unless you are running applications that use the message or semaphore features.
| Home |
|---|
sar -q reports the average queue length while the queue is occupied, and percent of time occupied.
An example of sar -q follows:
11:00:56 runq-sz %runocc swpq-sz %swpocc
11:01:07 1.7 98 1.5 36
11:01:17 1.0 63 1.0 31
11:01:27 1.0 58 1.0 49
Average 1.3 74 1.2 39 |
In this example, the processor utilization (%runocc) varies between 58% and 98%, while the fraction of time the swap queue is not empty (%swpocc) is 31% to 49%. This means that memory is not causing a major bottleneck in the system throughput, but more memory would help reduce the swapping activity.
The CPU utilization is listed by sar -u (default). At any given moment the processor is either busy or idle. When busy, the processor is in either user or system mode. When idle, the processor is waiting for input/output completion, waiting for memory or has no work to do.
The -u option of sar lists the percent of time that the processor is in system mode (%sys), user mode (%user), waiting for input/output completion (%wio), waiting for memory (%wmem) and idle time (%idle).
In typical timesharing use, %sys and %usr are about the same value. In special applications, either of these may be larger than the other without anything being abnormal. A high %wio generally means a disk bottleneck. A high %wmem, with degraded response time, may mean memory constraints.
An example of sar -u follows:
09:20:08 %usr %sys %wio %wmem %idle
09:40:12 6 7 2 0 86
10:00:03 7 9 3 0 80
10:20:07 14 16 10 0 61
Average 9 11 5 0 76 |
| Home |
|---|
sar -M reports the CPU usage condition for each processor. At any given moment, each processor is either busy or idle. When busy, the processor is in either user or system mode. The sar -M lists the time, the processor number (#CPU), the percentage of time that the processor is in system mode (%sys), user mode (%usr), idle time (%idle), and the percentage of vector instructions used (%vector).
An example of sar -M follows:
08:30:16 #CPU %usr %sys %idle %vector
08:30:21 0 94 5 1 89
1 99 1 0 90
08:30:26 0 90 6 4 83
1 99 1 0 94
Average 0 92 5 3 86
1 99 1 0 92 |
sar -v reports the status of process, inode, file, shared memory report, and shared memory file tables. From this report you know when the system tables need to be modified.
The values are given as level/table size. An example of sar -v follows:
17:36:05 proc-sz ov inod-sz ov file-sz ov lock-sz
17:36:35 17/ 40 0 39/ 80 0 29/ 80 0 0/ 50
17:37:05 19/ 40 0 46/ 80 0 35/ 80 0 0/ 50
17:37:35 18/ 40 0 43/ 80 0 34/ 80 0 0/ 50 |
This example shows that all tables are large enough to have no overflows. Sizes could be reduced to save main memory space if these are the highest values ever reported.
| Home |
|---|
sar -w reports swapping and switching activity. The following are some target values and observations.
An example of sar -w output follows:
19:53:44 swpin/s bswin/s swpot/s bswot/s pswch/s
19:53:58 0.0 0.0 0.0 0.0 37
19:54:14 0.0 0.0 0.0 0.0 39
19:54:24 0.0 0.0 0.0 0.0 39
Average 0.0 0.0 0.0 0.0 38 |
This example shows that there is sufficient memory for the currently active users, since no swapping occurs.
sar-W reports swapping activity separated small pages from large pages. The following are reported.
An example of sar -W output follows:
12:41:11 sswin/s sswot/s lswin/s lswot/s
12:42:11 0 0 0 0
12:43:12 0 0 1 1
12:43:12 0 1 1 2
Average 0 1 1 2 |
sar -w does not show which pages have insufficient (small or large) memory for the currently active users; sar -W does.
| Home |
|---|
sar -r reports the number of memory pages and swap file disk blocks that are currently unused. The following are reported.
An example of sar -r output follows:
12:01:51 freemem freeswap
12:56:52 208 5848 |
sar -y monitors terminal device activities. If you have a lot of terminal I/O, you can use this report to determine if there are any bad lines. Activities reported are defined as follows:
The number of modem interrupts per second (mdmin/s) should be close to 0, and the receive and transmit interrupts per second (xmtin/s and rcvin/s) should be less than or equal to the number of incoming or outgoing characters, respectively. If this is not the case, check for bad lines.
An example of sar -y output follows:
16:50:11 rawch/s canch/s outch/s rcvin/s xmtin/s mdmin/s
16:50:41 112 15 653 103 102 0
16:51:11 107 7 654 104 105 0
16:51:41 99 5 641 99 105 0
Average 106 9 649 102 104 0 |
| Home |
|---|
sar -d monitors device busy information. The following information is reported.
An example of sar -d output follows:
16:53:10 device %busy (r+w) counts/s
17:13:10 ID10 12 213
ID12 10 102
ID13 5 43
17:33:10 ID10 13 313
ID12 17 202
ID13 8 142
17:53:10 ID10 21 243
ID12 12 72
ID13 16 97
Average ID10 15 256
ID12 13 125
ID13 10 84 |
| Home |
|---|
sar -R reports large and small memory page usage of each RB.
For details on the RB, see Resource Block Facility. An example of sar -R output follows:
12:00:00 RB mem swap umem avail
12:10:00 L_page : 0 500 500 200 800
1 400 400 600 200
2 100 100 0 200
S_page : 0 1200 1400 380 2220
1 100 100 0 200
12:20:00 L_page : 0 500 500 200 800
1 400 400 600 200
2 100 100 0 200
S_page : 0 1200 1400 380 2220
2 100 100 0 200
Average L_page : 0 500 500 200 800
1 400 400 600 200
2 100 100 0 200
S_page : 0 1200 1400 380 2220
2 100 100 0 200
|
| Home |
|---|
| Contents | Previous Chapter | Next Chapter | Index |