linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* Linux client specweb test  hung
@ 2002-09-14 13:08 Hua Qin
  2002-09-14 13:27 ` Alan Cox
  0 siblings, 1 reply; 3+ messages in thread
From: Hua Qin @ 2002-09-14 13:08 UTC (permalink / raw)
  To: linux-kernel


Hi,

I think someone already have this discussion about this hung, but I did
not see some solutions about. Here is my test case:

1 Zeus web server: kernel 2.4.7-10
7 Specweb clients: kernel 2.2.16

After the client hang, I used netstat to see the connections, and found
some clients still keep the ESTABISHED connections, but there is no
connection on the server side.

I use tcpdump to look at how the server and clients to close the
connections. There are scenarios: 
1. The server send FIN first 
2. The client send  FIN first

If the server send  FIN first:
	client			server
	 <-----------FIN---------
	 ----------- ACK-------->
	 ----------- FIN-------->
	<------------ACK--------

If the client send the FIN first:

	client 			server
	--------------FIN-------->
	<-------------ACK--------
	
	or

	client			server
	-------------FIN--------->
	<------------FIN----------
	------------ACK---------->


The following is my guess:
When the iteration finishes everytime, if the FIN happeens to be sent from
the server side, the connection will closed correctly, if the FIN happens
to be sent from the client slide, the client will be hung. Is this
related to LAST-ACK issue?

Thanks!!
Hua


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Linux client specweb test  hung
  2002-09-14 13:08 Linux client specweb test hung Hua Qin
@ 2002-09-14 13:27 ` Alan Cox
  2002-09-15  1:08   ` Hua Qin
  0 siblings, 1 reply; 3+ messages in thread
From: Alan Cox @ 2002-09-14 13:27 UTC (permalink / raw)
  To: Hua Qin; +Cc: linux-kernel

On Sat, 2002-09-14 at 14:08, Hua Qin wrote:
> 
> Hi,
> 
> I think someone already have this discussion about this hung, but I did
> not see some solutions about. Here is my test case:
> 
> 1 Zeus web server: kernel 2.4.7-10
> 7 Specweb clients: kernel 2.2.16

2.4.17 is an old kernel with multiple errata fixes.
2.2.16 is obsolete, with known later networking, driver and other bugs
fixed.

Start with something current and see if you can duplicate the problem.


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Linux client specweb test  hung
  2002-09-14 13:27 ` Alan Cox
@ 2002-09-15  1:08   ` Hua Qin
  0 siblings, 0 replies; 3+ messages in thread
From: Hua Qin @ 2002-09-15  1:08 UTC (permalink / raw)
  To: Alan Cox; +Cc: linux-kernel


On 14 Sep 2002, Alan Cox wrote:

> On Sat, 2002-09-14 at 14:08, Hua Qin wrote:
> > 
> > Hi,
> > 
> > I think someone already have this discussion about this hung, but I did
> > not see some solutions about. Here is my test case:
> > 
> > 1 Zeus web server: kernel 2.4.7-10
> > 7 Specweb clients: kernel 2.2.16
> 
> 2.4.17 is an old kernel with multiple errata fixes.
> 2.2.16 is obsolete, with known later networking, driver and other bugs
> fixed.
> 
> Start with something current and see if you can duplicate the problem.
> 
> 
So do you think 2.4.7-10 or 2.4.17 has multiple errata fixes. or both :)
Thanks!
Hua


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2002-09-15  1:04 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-09-14 13:08 Linux client specweb test hung Hua Qin
2002-09-14 13:27 ` Alan Cox
2002-09-15  1:08   ` Hua Qin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).