[Fuego] 答复: bonnie fixup commit in Fuego pull request

Cai, Song cais.fnst at cn.fujitsu.com
Sat Sep 16 02:06:07 UTC 2017


I apologize for my so late reply.
I'm not familiar with bitbucket, I considered it would notice me the latest situation. 
And I was busy with another project, forgot the community email.

1. the first issue non-root:
If bonnie runs as root, it must add -u option.
As far as I know, almost all cases for agl runs as root. So I add the option "-u nobody" for the non-root bonnie case and change the ownership for working correctly.

2. parser.py
The original paser.py file missed some reference, so I add them and change the Random_Seek.Random_Seek to Random.seek by the way.

I would separate it and commit them again recently
Thanks for your suggestions.

--------------------------------------------------
B.R.
Cai Song
--------------------------------------------------




-----邮件原件-----
发件人: Bird, Timothy [mailto:Tim.Bird at sony.com] 
发送时间: 2017年9月8日 8:35
收件人: Daniel Sangorrin <daniel.sangorrin at toshiba.co.jp>; Cai, Song/蔡 嵩 <cais.fnst at cn.fujitsu.com>; fuego at lists.linuxfoundation.org
主题: RE: bonnie fixup commit in Fuego pull request



> -----Original Message-----
> From: Daniel Sangorrin [mailto:daniel.sangorrin at toshiba.co.jp]
> 
> > -----Original Message-----
> > From: Bird, Timothy [mailto:Tim.Bird at sony.com]
> > Sent: Wednesday, September 06, 2017 12:39 PM
> ...
> > For the second issue addressed by the commit:
> > In general it seems correct to remove the latency measures, since 
> > they are
> not in the testlog output that I can see, and they certainly
> > are not parsed by the parser for this Benchmark.  The name change 
> > from
> Random_Seek.Random_Seek to Random.Seek is also
> > appropriate, but there needs to be a change to the name in the
> measurements dictionary in parser.py to match.
> 
> Sorry, I haven't been able to review the patch. I just wanted to say 
> that certain measures will or will not be in the log depending on the 
> spec but also on the board itself.

I figured this might be the case - that the latencies were not in the test log results that I saw.

However, the bonnie parser.py doesn't have code to parse any latencies at all from the bonnie test log.  Perhaps this was intended as an extension to the bonnie parser in the future?

In any event, since they can never appear in a run.json file with the current parser, I think the right thing is to remove the latency fields from the reference.json file for now.  If we change the parser in the future, we can modify the reference.json file to match.
 -- Tim







More information about the Fuego mailing list