DSys RG A̶u̶g̶ Early Sept 2022: FLP Impossibility (1985)

Paper: https://groups.csail.mit.edu/tds/papers/Lynch/jacm85.pdf

00:28:05	Marc-Antoine Parent:	raft has timers, so not asynchronous
00:30:50	Michael Mure:	for CRDTs, isn't that that there is causality in messages, which bounds the time and avoid the full asynchronous problem?
00:31:39	Uchenna Nwanyanwu:	Sorry about that, I didn't know that the mic was enabled
00:31:45	Brooklyn Zelenka (@expede):	No prob!
00:35:00	Marc-Antoine Parent:	I'd say it's a different form of consensus
00:35:17	Marc-Antoine Parent:	consensus without awareness of consensus...
00:38:02	Brian:	https://www.podc.org/influential/2001-influential-paper/
00:38:54	Eleanor:	thank you for the link!
00:39:17	Jagan:	What is the relation between TCP and consensus?
00:40:30	Kim:	There are cases where both sides agree that delivery happened, at least.
00:41:08	Mauve:	(sorry was there supposed to be a IPVM community call in here in 7 mins?)
00:41:29	Steve Moyer:	seq is also useful in TCP if you happen to get a packet twice
00:41:35	Brooklyn Zelenka (@expede):	@mauve tomorrow!
00:41:50	James Walker (@walkah):	I think tomorrow is WNFS? IPVM in my calendar isn't until the 20th
00:42:20	Mauve:	I think my calendar is messed up. Apologies for the intrusion. 😁
00:42:34	Sodium :	\o
00:42:35	James Walker (@walkah):	You're more than welcome to stick around :)
00:55:16	Marc-Antoine Parent:	https://en.wikipedia.org/wiki/Network_Time_Protocol#Clock_synchronization_algorithm
00:55:52	Marc-Antoine Parent:	there is a feedback loop
00:59:54	Sodium :	"What’s the Difference? Efficient Set Reconciliation without Prior Context" https://www.ics.uci.edu/~eppstein/pubs/EppGooUye-SIGCOMM-11.pdf
01:00:00	Eleanor:	very interested in next months paper
01:04:49	Jon Forsyth:	Does Byzantine have a formal definition in Distributed Systems?
01:04:55	Marc-Antoine Parent:	If so also look at https://github.com/Libero0809/PBS.git
01:05:07	Marc-Antoine Parent:	(@Eleanor)
01:05:28	Eleanor:	thanks
01:05:50	Sodium :	Lamport's "The Byzantine Generals Problem": https://lamport.azurewebsites.net/pubs/byz.pdf
01:06:15	Jon Forsyth:	thx
01:06:18	Marc-Antoine Parent:	(I think we should look at this one too personally!)
01:07:19	Marc-Antoine Parent:	rho? do you mean sigma?
01:10:18	Nithin Thomas:	the last section briefly discussed consensus when the number of dead processes are known. it sounds really similar to RAFT. I am curious about that in the sense if asynchronous communication can be used to setup something like RAFT?
01:11:34	Marc-Antoine Parent:	Nithin: interesting question! But... would you not say that raft uses timers to add synchronicity to an asynchronous basis?
01:14:26	Nithin Thomas:	yeah that makes sense, thank you
01:15:22	Daniel Holmgren:	have another meeting coming up so gotta run. thanks yall!
01:15:25	Marc-Antoine Parent:	Shannon's theory of communication
01:15:31	Brooklyn Zelenka (@expede):	Thanks Daniel!
01:16:11	Steve Moyer:	was thinking of Shannon while we were discussing NTP
01:17:36	Sodium :	yes, the messages are addressed to individual processes
01:18:20	Sodium :	"Processes communicate by sending each other messages. A message is a pair (p, m), where p is the name of the destination process and m is a “message value” from a fixed universe M. "