Remove hard code mapping in dstat tool, Resolves git issue #2017

[lmchilton]

Updated pcp dstat tool to not rely on hard
coded mappings between instance filtering
command line options and the specific plugin
they apply to. Updated QA test #1207 and the
pcp-dstat(5) man page.

Linked to PR #2196

[wcohen]

The patch is looking really close. One issue I saw when trying out the branch with the tests is that 1187 fails with the patch. It looks like the following line in the test is failing because the -D sda,total isn't being handled correctly:

dstat "Per Disk" --disk -D sda,total --disk-wait

For pcp-dstat.5 it would be good to describe the mapping of cpu, disk, etc. to the -C, -D, etc. like what is done for printtype.

[wcohen]

One of the things that bothered me was that qa test 1187 failed when this patch was installed. However, it appears to be an issue with the test itself. The old hardcoded dstat would always prints each disk metrics for disk-wait plugin regardless of what was selected with the -D option. The qa1187 test is trying to get data for an disk that isn't in the data, sda. There is a vda. Would suggest an additional patch that correct the qa 1187 to provide more realistic test:

dstat "Per Disk" --disk -D sda,total --disk-wait

And adjust the output to be the expected:

----system---- --dsk/vda----dsk/total- ---vda---
     time     | read  writ: read  writ|rawa wawa
10-02 17:04:49|           :           |         
10-02 17:04:50|   0   235k:   0   235k|   0 0.00
10-02 17:04:51|   0     0 :   0     0 |   0    0
10-02 17:04:52|   0     0 :   0     0 |   0    0

At this point the patch appears to be working as expected. So once there are patches for pcp-dstat.5 documentation and the qa/1187 test to function correctly. This should be merged in.

[wcohen]

I have put patches for the update to the pcp-dstat.5 and qa/1187 on https://github.com/wcohen/pcp/tree/wcohen/pr2218

[wcohen]

I am comfortable that the patch is reasonable at this point. When experimenting with the patch for support the GPU plugin I addressed a number of other issues and place the patches on https://github.com/wcohen/pcp/tree/wcohen/

• A couple of the tests qa/1187 and qa1207 were not quite right
• dstat would always set the default grouptype to 2 even if the grouptype in the plugin was something else
• A few of the plugins needs to specify the default grouptype