Hi Dave, From: David Miller <[EMAIL PROTECTED]> Subject: Re: [PATCH 2.6.22] TCP: Make TCP_RTO_MAX a variable (take 2) Date: Thu, 12 Jul 2007 13:24:48 -0700 (PDT)
> From: OBATA Noboru <[EMAIL PROTECTED]> > Date: Thu, 12 Jul 2007 22:59:50 +0900 (JST) > > > How do you think TCP timeouts in Linux can adapt to such changes > > in network environment? > > I'm honestly not interested in discussing this any more > and Ian has even showed that the RFCs state that if we have > a maximum it must be at least 60. > > So really, there is no chance of merging a TCP_RTO_MAX > decreasing patch, sorry. My apologies for replying this late as I have no idea what I can do next at this point. But let me throw away my TCP_RTO_MAX decreasing patch and step back to the original problem to continue discussion. The original problem was: * A TCP packet is delayed upon a physical failure of a failover-and-recovery-capable network, until the failure is detected and recovered. The RTO of packet grows exponentially (say, up to 10 or 15 seconds) during failure detection and recovery. * While the application expects the TCP packet to be sent just after the recovery, but it may not necessarily be retransmitted immediately because of a large RTO. * Having this large RTO, the application TCP timer can expire before the retransmission packet is sent. The application timeout is considered as a major failure and is desirable to be avoided. And your first reply was: From: David Miller <[EMAIL PROTECTED]> Subject: Re: [PATCH 2.6.22] TCP: Make TCP_RTO_MAX a variable (take 2) Date: Thu, 12 Jul 2007 02:37:10 -0700 (PDT) > TCP's timeouts are perfectly fine, and the only thing you > might be showing above is that the application timeouts > are too short or that TCP needs notifications. Is it correct that you think my problem can be addressed either by the followings? (1) Make the application timeouts longer. (Steve has shown that making an application timeouts twice the failover detection timeout would be a solution.) (2) Let TCP have a notification of some kind. If so, I would like to explore the way to implement it. I am interested in TCP notification now because making the application timeouts longer in my case is just not feasible. Regards, -- OBATA Noboru ([EMAIL PROTECTED]) - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html