Netscape consumes cpu after POST

Netscape consumes cpu after POST

Post by Stefan Andersso » Wed, 20 Sep 2000 04:00:00



I have a problem that Netscape after submitting a form having the POST
method, eats cpu until getting the reply (or Stop is pushed). This is a
particular problem if connected to a webserver running on the same
computer since Netscape steals all cpu-time from the webserver.
Strangely this seems to happen only if the form has the post method, not
GET.

The problem occurs with 4.7 and 4.5 on both NT4 and Win2000.

Anyone have a clue?

--
Stefan Andersson
Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

 
 
 

Netscape consumes cpu after POST

Post by Urban Rüeg » Wed, 20 Sep 2000 04:00:00


do you hav between <form> and </form> an <input type = "texte...> or <input
type="submit"...> tag?

show whole source-code, probably I can help, I had lots of Netscape-troubles
to solve,

Urban


Quote:> I have a problem that Netscape after submitting a form having the POST
> method, eats cpu until getting the reply (or Stop is pushed). This is a
> particular problem if connected to a webserver running on the same
> computer since Netscape steals all cpu-time from the webserver.
> Strangely this seems to happen only if the form has the post method, not
> GET.

> The problem occurs with 4.7 and 4.5 on both NT4 and Win2000.

> Anyone have a clue?

> --
> Stefan Andersson
> Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> Tel: +46 8 6900750, +46 18 301273, +46 70 3710667


 
 
 

Netscape consumes cpu after POST

Post by Stefan Andersso » Thu, 21 Sep 2000 04:00:00


This show up on any page having a form with post. I wrote a simple asp-page to
demo the problem (attached). Running this through the local IIS (ms www-server),
it takes a few secs to get ther reply if I use IE. The delay loop make the
server consume some cpu time befor replying. However accessing the page with NS,
it takes a minutes. Watching with the windows taskmanager, I can see that NS,
after submitting the form keeps consuming almost all available cpu time, thus
not giving the server any chance to do its work.
NS behaves the same way accessing a non-local server, although the reply-time
is, naturally, not affected.


> do you hav between <form> and </form> an <input type = "texte...> or <input
> type="submit"...> tag?

> show whole source-code, probably I can help, I had lots of Netscape-troubles
> to solve,

> Urban



> > I have a problem that Netscape after submitting a form having the POST
> > method, eats cpu until getting the reply (or Stop is pushed). This is a
> > particular problem if connected to a webserver running on the same
> > computer since Netscape steals all cpu-time from the webserver.
> > Strangely this seems to happen only if the form has the post method, not
> > GET.

> > The problem occurs with 4.7 and 4.5 on both NT4 and Win2000.

> > Anyone have a clue?

> > --
> > Stefan Andersson
> > Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> > Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

--
Stefan Andersson
Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
Tel: +46 8 6900750, +46 18 301273, +46 70 3710667
 
 
 

Netscape consumes cpu after POST

Post by Stefan Andersso » Thu, 21 Sep 2000 04:00:00


Oops, forgot to attach.


> This show up on any page having a form with post. I wrote a simple asp-page to
> demo the problem (attached).


<%
 Option Explicit
' response.buffer =true
%>
<html>
<head>
</head>
<body>

<%
response.write "hej<br>" & vbcrlf
'response.flush
dim i
dim f
f= 1.001
for i = 1 to 1000000
 f = f * 1.00002
next
dim objCtl
for each objCtl in Request.Form
   response.write objCtl & ": " & Request.Form(objCtl) & vbcrlf
next
%>

<form action="nstest1.asp" method="post" name="tf">
<input type=checkbox name=cb1>
<p>
<input type="text" name="tx1" value="broccoli">
<p>
Self-submitting:
<input type=checkbox name=cb2 onclick="document.tf.submit()" >
<p>
<input type=submit value=Submit>
</form>

</body>
</html>

--
Stefan Andersson
Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

 
 
 

Netscape consumes cpu after POST

Post by Urban Rüeg » Fri, 22 Sep 2000 04:00:00


try:

response.buffer =false

if response.buffer = true, then the browser will wait until the end of the 1
million! calculations to show the whole page.
IE acts much diffrent..
lets see, if it works.

urban


> Oops, forgot to attach.


> > This show up on any page having a form with post. I wrote a simple
asp-page to
> > demo the problem (attached).


> <%
>  Option Explicit
> ' response.buffer =true
> %>
> <html>
> <head>
> </head>
> <body>

> <%
> response.write "hej<br>" & vbcrlf
> 'response.flush
> dim i
> dim f
> f= 1.001
> for i = 1 to 1000000
>  f = f * 1.00002
> next
> dim objCtl
> for each objCtl in Request.Form
>    response.write objCtl & ": " & Request.Form(objCtl) & vbcrlf
> next
> %>

> <form action="nstest1.asp" method="post" name="tf">
> <input type=checkbox name=cb1>
> <p>
> <input type="text" name="tx1" value="broccoli">
> <p>
> Self-submitting:
> <input type=checkbox name=cb2 onclick="document.tf.submit()" >
> <p>
> <input type=submit value=Submit>
> </form>

> </body>
> </html>

> --
> Stefan Andersson
> Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

 
 
 

Netscape consumes cpu after POST

Post by Stefan Andersso » Fri, 22 Sep 2000 04:00:00


I've tried it. It makes a difference in that the part of the page generated
before the loop shows up earlier, but the problem is left. Netscape keeps
consuming cpu until it has received the whole page.


> try:

> response.buffer =false

> if response.buffer = true, then the browser will wait until the end of the 1
> million! calculations to show the whole page.
> IE acts much diffrent..
> lets see, if it works.

> urban



> > Oops, forgot to attach.


> > > This show up on any page having a form with post. I wrote a simple
> asp-page to
> > > demo the problem (attached).


> > <%
> >  Option Explicit
> > ' response.buffer =true
> > %>
> > <html>
> > <head>
> > </head>
> > <body>

> > <%
> > response.write "hej<br>" & vbcrlf
> > 'response.flush
> > dim i
> > dim f
> > f= 1.001
> > for i = 1 to 1000000
> >  f = f * 1.00002
> > next
> > dim objCtl
> > for each objCtl in Request.Form
> >    response.write objCtl & ": " & Request.Form(objCtl) & vbcrlf
> > next
> > %>

> > <form action="nstest1.asp" method="post" name="tf">
> > <input type=checkbox name=cb1>
> > <p>
> > <input type="text" name="tx1" value="broccoli">
> > <p>
> > Self-submitting:
> > <input type=checkbox name=cb2 onclick="document.tf.submit()" >
> > <p>
> > <input type=submit value=Submit>
> > </form>

> > </body>
> > </html>

> > --
> > Stefan Andersson
> > Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> > Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

--
Stefan Andersson
Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
Tel: +46 8 6900750, +46 18 301273, +46 70 3710667
 
 
 

Netscape consumes cpu after POST

Post by Urban Rüeg » Fri, 22 Sep 2000 04:00:00


Now, I'm wondering about the diffrence between the browsers and the get/post
method.
Can you replicate those   mesurements?
Because the script is running with both browsers on the server! and
consuming sertime anyway. Then the multitasking is giving time to every
task. Though, for me it's now very hard to believe, that Netscape should get
a higher priority from the System than IE???
Consuming CPU-Time, does that really mean, that the CPU has no capacity for
other tasks? Or does it just mean, that you wait that Netscape shows the
whole page???

For that case thats normal,  that the netscape rendering machine waits much
longer than IE, vory often until the </body>-tag.
Why your script has to work so long? Is this test real-like? do you really
have 1 million calculations like this? In a so much slow language? If yes,
you could make the client make calculations, with javascript per example...
Anyway it's never a good idea to have scripts running so long time on
servers...

Urban


> I've tried it. It makes a difference in that the part of the page
generated
> before the loop shows up earlier, but the problem is left. Netscape keeps
> consuming cpu until it has received the whole page.


> > try:

> > response.buffer =false

> > if response.buffer = true, then the browser will wait until the end of
the 1
> > million! calculations to show the whole page.
> > IE acts much diffrent..
> > lets see, if it works.

> > urban



> > > Oops, forgot to attach.


> > > > This show up on any page having a form with post. I wrote a simple
> > asp-page to
> > > > demo the problem (attached).


> > > <%
> > >  Option Explicit
> > > ' response.buffer =true
> > > %>
> > > <html>
> > > <head>
> > > </head>
> > > <body>

> > > <%
> > > response.write "hej<br>" & vbcrlf
> > > 'response.flush
> > > dim i
> > > dim f
> > > f= 1.001
> > > for i = 1 to 1000000
> > >  f = f * 1.00002
> > > next
> > > dim objCtl
> > > for each objCtl in Request.Form
> > >    response.write objCtl & ": " & Request.Form(objCtl) & vbcrlf
> > > next
> > > %>

> > > <form action="nstest1.asp" method="post" name="tf">
> > > <input type=checkbox name=cb1>
> > > <p>
> > > <input type="text" name="tx1" value="broccoli">
> > > <p>
> > > Self-submitting:
> > > <input type=checkbox name=cb2 onclick="document.tf.submit()" >
> > > <p>
> > > <input type=submit value=Submit>
> > > </form>

> > > </body>
> > > </html>

> > > --
> > > Stefan Andersson
> > > Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> > > Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

> --
> Stefan Andersson
> Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

 
 
 

Netscape consumes cpu after POST

Post by Jason » Tue, 26 Sep 2000 12:21:01


I have the exact same program myself and what i do to work around the
problem is to set the priority of the netscape.exe process to low or
even idle before doing any of my testing and it works fine.

To me it seems like netscape is polling winsock for the response instead
of using an event based method which they seem to be doing quite fine
for GET requests. Yet another reason why I do not use Netscape as my
primary browser.

Jason


> I have a problem that Netscape after submitting a form having the POST
> method, eats cpu until getting the reply (or Stop is pushed). This is a
> particular problem if connected to a webserver running on the same
> computer since Netscape steals all cpu-time from the webserver.
> Strangely this seems to happen only if the form has the post method, not
> GET.

> The problem occurs with 4.7 and 4.5 on both NT4 and Win2000.

> Anyone have a clue?

> --
> Stefan Andersson
> Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
> Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

 
 
 

Netscape consumes cpu after POST

Post by Stefan Andersso » Tue, 26 Sep 2000 04:00:00



> Now, I'm wondering about the diffrence between the browsers and the get/post
> method.
> Can you replicate those   mesurements?

Sure.

Quote:

> Because the script is running with both browsers on the server! and
> consuming sertime anyway. Then the multitasking is giving time to every
> task. Though, for me it's now very hard to believe, that Netscape should get
> a higher priority from the System than IE???
> Consuming CPU-Time, does that really mean, that the CPU has no capacity for
> other tasks? Or does it just mean, that you wait that Netscape shows the
> whole page???

As Jason W points out, it makes a big difference if you set Netscapes priority
to "lower than normal". Then the server is given time to finish its job despite
Netscape trying to "steal" cpu-time.

Quote:

> For that case thats normal,  that the netscape rendering machine waits much
> longer than IE, vory often until the </body>-tag.
> Why your script has to work so long? Is this test real-like? do you really
> have 1 million calculations like this? In a so much slow language?

For example, a jsp (Java Server Pages) page, is compiled the first time it is
accessed, and that may take considerable amount of work.

Quote:> If yes,
> you could make the client make calculations, with javascript per example...
> Anyway it's never a good idea to have scripts running so long time on
> servers...

This might be wise in general. However there are many applications requiring
quite heavy server-side execution.

--
Stefan Andersson
Tacton Systems AB, Stockholm, Sweden  http://www.tacton.com
Tel: +46 8 6900750, +46 18 301273, +46 70 3710667

 
 
 

1. Netscape consumes the cpu completely on WinNT

Hi,
If this was diskussed before forward > /dev/nul

Netscape seems to use about 70..100% cpu time to connect or load data
from the net.
We have been experienced this since using NT 4.0 Beta and Beta 2 but
checking against
version 3.51 showed the same there (it is just not as easily to notice
because you have
to use permormance meter to be sure). This problem is even there if you
download a file
using netscape.

Naturally using a standalone client it is difficult to meassure if there
is
any additional cpu load at all.

We are using Netscape Navigator 3.0b4 and b5.

Any guess ?
Cheers
Christian
--
Christian Mallwitz NetConsult Communications Germany
www: http://www.imn.htwk-leipzig.de/~mallwitz
pgp: 02 30 7E 1A 7A C1 5C 16 2B 4D 53 A9 1B 05 F7 DD

2. Start, Restart delayed

3. VB app with Windows Media 9 control consuming 100% CPU

4. Removal of Critical Update Notification

5. Encoder is CPU consuming, hardware encode can be used?

6. wxDbTable::WX_ROW_MODE_QUERY

7. viewing input consumes 100% CPU when encoding miniDV via IEEE 1394 with Windows Media Encoder 7

8. Newbie needs XP update help!

9. IE6 Won't close after running SPYBOT and consumes 50%cpu

10. SVCHOST consuming all system memory and CPU time

11. System Process Consumes CPU Resources

12. SVCHOST.exe consuming CPU and memory

13. Repost: svchost.exe consumes CPU when file system accessed