ziorep_utilization examples

Examples demonstrate the summary option, the end range option, and the interval option.

This example shows how the summary option lists the date ranges of the collected data, its interval length, and the involved hardware:

# ./ziorep_utilization -s multipath_stress.log

Data Summary
------------
Aggregated range: 2008-11-13 08:56:49 to 2008-11-13 16:12:53
Detailed range:   2008-11-13 16:12:57 to 2008-11-13 20:56:45
Interval length:  4 seconds
HBA/CHPID:        0.0.3c40/52
                  0.0.3c00/50
WWPN/LUN (dev):   500507630313c562/4013401500000000 (/dev/sda)
                  500507630303c562/4013401500000000 (/dev/sdb)
                  500507630313c562/4013401400000000 (/dev/sdc)
                  500507630303c562/4013401400000000 (/dev/sdd)
                  500507630313c562/4013401a00000000 (/dev/sde)
                  500507630303c562/4013401a00000000 (/dev/sdf)
                  500507630313c562/4013401c00000000 (/dev/sdg)
                  500507630303c562/4013401c00000000 (/dev/sdh)
                  500507630313c562/4013401800000000 (/dev/sdi)
                  500507630303c562/4013401800000000 (/dev/sdj)
                  500507630313c562/4013401b00000000 (/dev/sdk)
                  500507630303c562/4013401b00000000 (/dev/sdl)
                  500507630313c562/4013401700000000 (/dev/sdm)
                  500507630303c562/4013401700000000 (/dev/sdn)

This example shows the output from an input file containing data for two FCP channels with one FCP device hosted on each:

# ./ziorep_utilization multipath_stress -e "2008-11-13 16:13:09"
CHP|adapter in %-|--bus in %---|--cpu in %---|
ID min max    avg min max  avg min max avg
2008-11-13 16:12:53 Aggregated Frame
52  0  57     2.4 2   53  22.4 2   15  5.1
50  0  59     2.5 2   52  22.4 2   15  5.1
16:12:57
52  9    9     9.0 29  29  29.0 4   4  4.0
50 12  12    12.0 28  28  28.0 3    3  3.0
16:13:01
52  1    1     1.0 24  24  24.0 3   3  3.0
50  1    1     1.0 29  29  29.0 4   4  4.0
16:13:05
52 10  10     10.0 25  25  25.0 3   3  3.0
50  4   4      4.0 25  25  25.0 3   3  3.0
...
2008-11-14 00:00:01
...
CHP Bus-ID  |--qdio util.i.%--|queu|fail|-thp / MB/s--|I/O reqs-|
 ID            min   max   avg full erc      rd    wrt  rd  wrt
2008-11-13 16:12:53 Aggregated Frame
 50/0.0.3c00   0.0 100.0  96.7   28K  0   16.5    5.8 2.0M 455K
 52/0.0.3c40   0.0 100.0  96.6   28K  0   15.5    5.0 2.0M 463K
16:12:57
 50/0.0.3c00   0.0 100.0  97.2     0  0   10.4    6.2 4.4K 812
 52/0.0.3c40   0.0 100.0  96.8     0  0    8.1    6.4 5.2K 894
16:13:01
 50/0.0.3c00   0.0 100.0  97.3     0  0    9.9   12.1 3.5K 248
 52/0.0.3c40   0.0 100.0  97.7     0  0   10.1   14.5 2.5K 175
16:13:05
 50/0.0.3c00   0.0 100.0  98.5     0  0    8.2    7.2 3.5K 116
 52/0.0.3c40   0.0 100.0  98.0     0  0   10.3    8.1 3.7K 113
...
2008-11-14 00:00:01
...

Note that numbers can be abbreviated if space does not suffice. For example, 17 361 can be abbreviated to 17K.

The output comes in two parts: The first part gives the utilization of the whole FCP channels, while the second part gives data for all FCP devices.

The meaning of the columns is as follows:

CHPID
The channel path ID of the device.
Bus-ID
The eight-character device bus ID of the FCP device.
adapter, bus, and cpu
The FCP channel, bus, and CPU utilizations as reported by the FCP channel statistics in percent. For example, a value of 37.2 represents a value of 37.2 percent.
fail erc
The number of error recovery conditions.
qdio utilization
The min, max and avg columns give the minimum, maximum and average outbound queue utilization respectively.
queu full
The number of instances where a request to the FCP device could not be submitted due to no empty slots left in the outbound queue.
thp / MB/s
This is the average throughput over time (volume transmitted / elapsed time) in megabytes per second, not over number of requests (sum over all request throughputs / number of requests)!

This means that a long-running request with a significantly different throughput profile from the rest will have a bigger impact than a brief one with the same throughput profile would. This gives a better impression of the overall profile and especially makes requests with very low throughputs have a bigger impact, making it easier to detect anomalies.

The abbreviations rd and wrt mean read and write throughput.

I/O reqs
is the number of I/O requests processed in the interval.

The abbreviations rd and wrt mean read and write requests.

Each new day and each new interval are marked by a line. All applicable FCP channels are then listed on individual lines for each timeslot. The label Aggregated highlights ranges in the data where the source data was already aggregated and hence cannot be processed further. If you select a timeframe that touches the range in which only aggregated data is available, the complete aggregated data will be reprinted. However, this can only be at most one dataset per device, and will only appear as the first line in the output.

In this example an interval length of 0 is chosen, causing all data in the specified timeframe to be aggregated into a single entry:

# ./ziorep_utilization multipath_stress.log -i 0
CHP|adapter in %-|--bus in %---|--cpu in %---|
 ID min max   avg min max   avg min max   avg
2008-11-13 20:56:45
 52   0  57   1.4  2  53   22.3   2  15   5.0
 50   0  59   1.5  2  52   22.3   2  15   5.0
CHP Bus-ID  |--qdio util.i.%--|queu|fail|-thp / MB/s--|I/O reqs-|
 ID           min    max   avg full  erc   rd      wrt   rd  wrt
2008-11-13 20:56:45
 50/0.0.3c00  0.0  100.0  96.7  30K    0 16.5      5.8 2.0M 455K
 52/0.0.3c40  0.0  100.0  96.6  31K    0 15.5      5.0 2.0M 463K