Hi Mitchell, Agreed, I rather not have a TCP style in OSPF graceful restart.
But I am talking about TCP Timeout, but an application has already done much work to send data to end point, before time out occurs. I would rather have a request reply mechanism , which is similar to pinging end points, before actually beginning the signaling. Which I have shown in the time line. Stressing the point of only what I want before the irreversible restart. Regards, Abhay ----- Original Message ----- From: "Erblichs" <[EMAIL PROTECTED]> To: "Abhay D.S" <[EMAIL PROTECTED]> Cc: "Acee Lindem" <[EMAIL PROTECTED]>; <[email protected]> Sent: Saturday, October 20, 2007 4:02 PM Subject: Re: [OSPF] Cascaded Graceful Restarts with NSF for OSPF protocolQuestion > Abhay D.S., > > Sorry about the top post, but.. > > TCP has a slow-start mechanism with ACK > clocking to verify completion of xmit'ed data, > until a threshold is met or, > > Until 2 or 3 DUP ACKs are seen and then > rexmited or a coarse grain timeout occurs, > > However, TCP does not guarantee that the > recvr will actually recieve data xmited > by the send, just that if the data arrives > then, it will send acks.. > > And again, their is no guarantee that the > acks will be recieved, so the sender can > send more data. > > Mitchell Erblich > ----------------- > > > "Abhay D.S" wrote: > > > > Hello Acee, > > > > a) Restart complete variable poll is one method after restart > > > > I am trying to find a method before restart to be deterministic. > > > > A simple analogy in TCP. > > > > For example..in TCP, just after connection establishment, I want > > start sending data, at the same time the remote link had some problem. > > > > > > a) Scenario 1 > > > > 0sec .....5 Sec...(ok).....10 Sec (ok).....15 Sec (ok).....[DONE] > > > > Continue restart y/n ? : y > > > > Attempting graceful restart..... > > > > [OK] > > > > b) Scenario 2 > > > > 0sec .....5 Sec...(ok).....10 Sec (Wait a While)....[Retrying due to > > network conditions] > > > > > > 0sec .....5 Sec...(Wait a While).....10 Sec (Ok)...15 Sec > > (ok).....[DONE] > > > > Continue restart y/n ? : y > > > > Attempting graceful restart...... > > > > [OK] > > > > c) Scenario 3 > > > > 0sec .....5 Sec...(ok).....10 Sec (Wait a While)....[Retrying due to > > network conditions] > > > > > > 0sec .....5 Sec...(Wait a While).....10 Sec (Ok)...15 Sec (Wait a > > While).....[Retrying due to network conditions] > > > > 0sec .....5 Sec...(ok).....10 Sec (Wait a While)....[Retrying due to > > network conditions] > > > > Continuing restart may affect forwarding... > > Continue anyway ? y/n .....n > > > > [OK] > > > > > > The time line above considers a network stable. > > > > > > After the restart is done I can follow method a) > > > > > > Thanks for your reply, > > Regards, > > Abhay > > > > > > > > > > > > > > ----- Original Message ----- > > From: Acee Lindem > > To: Abhay D.S > > Cc: [email protected] > > Sent: Friday, October 19, 2007 9:13 PM > > Subject: Re: [OSPF] Cascaded Graceful Restarts with NSF for > > OSPF protocol Question > > > > Hi Abhay, > > Most of this borders on the implementation of the networking > > equipment's in-service upgrade. For now, let's just assume > > that you have redundant data place connectivity for the new > > line cards and solely need do the graceful restarts as > > indicated below. Again, this borders more on implementation. > > However, now the OSPFv2 MIB as described in RFC 4750 does > > support a variable which could be polled to determine when > > restart on a given router has completed. > > > > ospfRestartStatus OBJECT-TYPE > > SYNTAX INTEGER { notRestarting (1), > > plannedRestart (2), > > unplannedRestart (3) > > } > > MAX-ACCESS read-only > > STATUS current > > DESCRIPTION > > "Current status of OSPF graceful restart." > > ::= { ospfGeneralGroup 21 } > > > > > > Hope this helps, > > Acee > > > > On Oct 18, 2007, at 11:56 PM, Abhay D.S wrote: > > > > > hi all, > > > > > > Suppose I am providing real-time transactions over IP in a > > > non NSR environment(I dont invest much on getting a backup > > > RP i.e). > > > > > > But I have NSF, owing to the ease of which OSPF can be > > > used, I > > > use OSPF to manage connections to WAN. > > > > > > And I have bought new I/0 modules which require a restart > > > for > > > all my routers for proper installation. > > > > > > Luckily, I have inservice upgrade on the boxes. > > > > > > Assuming that I am a new operator without any > > > certifications under > > > my belt, I dont have a beautiful NMS with colored buttons. > > > > > > But I must implement the new I/O cards and its features > > > without disturbing > > > the users who are using the network for routing > > > transactions such as data > > > backup over network or voice forwarding and other > > > distributed applications > > > running over TCP/IP using my network. > > > > > > I want to graceful restart all my routers one by one, > > > called as cascaded > > > re-starts, how can I make sure the transition is done > > > seamlessly. > > > > > > I want the "this thing" to be automatic, since each router > > > is graceful > > > restarted > > > and then then the next one is re-started and then the next > > > one, > > > automatically. > > > > > > I can write a script to do that, but I am not sure the > > > forwarding will be > > > safe, > > > because it is difficult to control the network transient > > > conditions. > > > > > > Any mechanism available ?. > > > > > > Thanks and Regards, > > > Abhay > > > > > > > > > > > > > > > _______________________________________________ > > > OSPF mailing list > > > [email protected] > > > https://www1.ietf.org/mailman/listinfo/ospf > > > > > > > --------------------------------------------------------------- > > _______________________________________________ > > OSPF mailing list > > [email protected] > > https://www1.ietf.org/mailman/listinfo/ospf _______________________________________________ OSPF mailing list [email protected] https://www1.ietf.org/mailman/listinfo/ospf
