Re: wrong phase duration in WPG report summary

--===============6274227008941146788==
Content-Type: multipart/alternative; boundary=001a113eccda548b23052d1ee6ef

--001a113eccda548b23052d1ee6ef
Content-Type: text/plain; charset=UTF-8

Hi Alex,

Thank you so much for the detailed explanation. Using --phases option in
reporter tool is very handy and useful. Thanks for the info :)

Regards,
Sanjeevi

On Thu, Mar 3, 2016 at 3:51 AM, Alex Rousskov <
[email protected]> wrote:

> On 03/02/2016 12:27 PM, Sanjeevi Rangan wrote:
>
> > I have done few tests using WPG client and server with multiple phases
> > (10min + 60min + 10min) and generated WPG report by feeding both client
> > and server logs to the reporter tool . During initial 10 mins the number
> > of concurrent connections would increase from 0 to max, the next 60 mins
> > would maintain the max number of concurrent connections and the in the
> > last 10 mins, the concurrent connection would decrease from max to 0. In
> > some tests, the executive summary table on the index page would report
> > duration as '10 min' and in some other tests, the duration value is
> > '1.00 hour' (60 mins, which is expected).
>
> This instability might be attributed to the actual peak request rates of
> the two phases being very close. Polygraph is just guessing which phase
> is more important to you...
>
>
> > If the duration is '10 min' then the other metrics data on the table
> > like response time and throughput would have values related to the
> > transactions that happened during that 10mins.
>
>
> > I would like to know what determines the phase duration in the report.
>
> The reported duration is the total measured duration of the phase(s)
> selected for the executive summary and the baseline report. What you
> probably want to know is which phase(s) are selected for the executive
> summary and the baseline report :-).
>
> By default, Polygraph currently selects the first phase with the highest
> measured request rate. There is a 1% fuzz that is used when comparing
> request rates, but when measured peak request rates are close, it may
> not be enough to produce stable guesses.
>
> Consider avoiding this uncertainty by configuring which PGL Phase(s) are
> the primary ones. Here is an excerpt from the v4.0.6 change.log:
>
> > - Added a PGL Phase.primary boolean field. If any of the phases
> >   have that field set, those phases and only those phases should
> >   be used for the executive summary and the baseline report. The
> >   new field is logged when phase stats are logged.
>
>
> You can also specify the primary phases during report generation, using
> the --phases command line parameter for the reporter tool. Naturally,
> the reporter parameter overwrites PGL configuration so you can produce
> multiple reports with different executive summaries using the same set
> of logs.
>
>
> HTH,
>
> Alex.
>
>


-- 
*Regards,*
*Sanjeevi Rangan*

--001a113eccda548b23052d1ee6ef
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div><div><div>Hi Alex,<br><br></div>Thank you so much for=
 the detailed explanation. Using --phases option in reporter tool is very h=
andy and useful. Thanks for the info :)<br><br></div>Regards,<br></div>Sanj=
eevi<br></div><div class=3D"gmail_extra"><br><div class=3D"gmail_quote">On =
Thu, Mar 3, 2016 at 3:51 AM, Alex Rousskov <span dir=3D"ltr">&lt;<a href=3D=
"mailto:[email protected]" target=3D"_blank">rousskov@measur=
ement-factory.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote=
" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><=
span class=3D"">On 03/02/2016 12:27 PM, Sanjeevi Rangan wrote:<br>
<br>
&gt; I have done few tests using WPG client and server with multiple phases=
<br>
&gt; (10min + 60min + 10min) and generated WPG report by feeding both clien=
t<br>
&gt; and server logs to the reporter tool . During initial 10 mins the numb=
er<br>
&gt; of concurrent connections would increase from 0 to max, the next 60 mi=
ns<br>
&gt; would maintain the max number of concurrent connections and the in the=
<br>
&gt; last 10 mins, the concurrent connection would decrease from max to 0. =
In<br>
&gt; some tests, the executive summary table on the index page would report=
<br>
&gt; duration as &#39;10 min&#39; and in some other tests, the duration val=
ue is<br>
&gt; &#39;1.00 hour&#39; (60 mins, which is expected).<br>
<br>
</span>This instability might be attributed to the actual peak request rate=
s of<br>
the two phases being very close. Polygraph is just guessing which phase<br>
is more important to you...<br>
<span class=3D""><br>
<br>
&gt; If the duration is &#39;10 min&#39; then the other metrics data on the=
 table<br>
&gt; like response time and throughput would have values related to the<br>
&gt; transactions that happened during that 10mins.<br>
<br>
<br>
&gt; I would like to know what determines the phase duration in the report.=
<br>
<br>
</span>The reported duration is the total measured duration of the phase(s)=
<br>
selected for the executive summary and the baseline report. What you<br>
probably want to know is which phase(s) are selected for the executive<br>
summary and the baseline report :-).<br>
<br>
By default, Polygraph currently selects the first phase with the highest<br=
>
measured request rate. There is a 1% fuzz that is used when comparing<br>
request rates, but when measured peak request rates are close, it may<br>
not be enough to produce stable guesses.<br>
<br>
Consider avoiding this uncertainty by configuring which PGL Phase(s) are<br=
>
the primary ones. Here is an excerpt from the v4.0.6 change.log:<br>
<br>
&gt; - Added a PGL Phase.primary boolean field. If any of the phases<br>
&gt;=C2=A0 =C2=A0have that field set, those phases and only those phases sh=
ould<br>
&gt;=C2=A0 =C2=A0be used for the executive summary and the baseline report.=
 The<br>
&gt;=C2=A0 =C2=A0new field is logged when phase stats are logged.<br>
<br>
<br>
You can also specify the primary phases during report generation, using<br>
the --phases command line parameter for the reporter tool. Naturally,<br>
the reporter parameter overwrites PGL configuration so you can produce<br>
multiple reports with different executive summaries using the same set<br>
of logs.<br>
<br>
<br>
HTH,<br>
<br>
Alex.<br>
<br>
</blockquote></div><br><br clear=3D"all"><br>-- <br><div class=3D"gmail_sig=
nature"><div dir=3D"ltr"><i><b>Regards,</b></i><div><i><b>Sanjeevi Ranga</b=
><b>n</b></i><i><b><br></b></i></div><br><div><br></div></div></div>
</div>

--001a113eccda548b23052d1ee6ef--

--===============6274227008941146788==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Users mailing list
[email protected]
http://lists.web-polygraph.org/mailman/listinfo/users
--===============6274227008941146788==--