LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Problems with IPVS

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Problems with IPVS
From: Roberto Nibali <ratz@xxxxxxxxxxxx>
Date: Fri, 10 Nov 2006 15:39:32 +0100
SE phone though acknowledges some previously acknowledged
packet (ethereal shows this as DupACK). web server then resends data but it does not pass through LVS-NAT.
We could enhance IPVS to accept such traffic, however I'd much rather like to understand the root of the problem.

Now I see such behaviour of IPVS as bogus. But I'm just an amateur. :)

I wouldn't classify the behaviour as "bogus", it's a design consideration. I've previously mentioned the possibility to handle the TCP state transition differently. Especially with the forward shared approach and the LVS-NAT we would have real TCP state tracking for session templates. LVS-DR and LVS-TUN will always be a compromise with regard to TCP state transition handling.

SE phone behaves strange but it does not violate TCP/IP. Or it does? But as

I don't know, there were not enough indications in your provided caps to qualify either statement.

you understand maybe only Vodafone could kick SE and make it program more
carefully. :)

Any they should, since SE is gaining momentum on the market. For some reason vodafone is busier figuring out how their SOA portfolio should look like for their potential customers than to actually address unimportant things like a buggy TCP stack. Either we find out directly what's wrong with those phones or we try to find an elegant solution for LVS, without breaking it for all the other people. The immediate fix I have in mind is too intrusive and severely lacks security, so we have to think about it a bit.

BTW, can you get the network event messages from the phone?

AT+CREG=1
AT*EAPP=3,1,"http://www.google.com";

Or we put a lwTCP stack on the phone :):

http://savannah.nongnu.org/projects/lwip/

As I've heard they did that to Nokia after one of new Nokia
phone behaved very badly. :)

Yeah, mobile phones are pushed out to the market so fast that quality assurance is almost not possible anymore. And often also not in the interest of the companies. You can charge your customers twice if you release a slightly defective product. This works in oligopolic markets where people depend on the product and where you have some sort of micro-billing and no concept of customer feedback integration.

Enjoy your weekend,
Roberto Nibali, ratz
--
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq' | dc

<Prev in Thread] Current Thread [Next in Thread>