These are chat archives for ethersphere/orange-lounge

3rd
Oct 2017
holisticode
@holisticode
Oct 03 2017 00:49
@lmars updated PR
@zelig ethersphere/simple-p2p-d3#20 (latest frontend PR)
holisticode
@holisticode
Oct 03 2017 02:34
@zelig Network-testting-and-simulation-framework status updated at https://github.com/ethersphere/swarm/wiki/Working-groups#network-testing-and-simulation-framework
After the demo this morning it is my conclusion that the current implementation of the simulation frontend is a successful PoC, in that it has shown that we can visualise a network in realtime, and query certain parameters via a HTTP API
Nevertheless, it has also become clear that we can improve its usefulness - it's of no use to have great eye-candy but little extracted information or insights
to this end, I believe we need to identify, track, and document the requirements for such a visualisation frontend in a structured way
@/all everyone invited to contribute to requirements - ideally it would become a tool at the disposal of everyone
Viktor Trón
@zelig
Oct 03 2017 06:46
@holisticode :clap: very cool
orenyodfat
@orenyodfat
Oct 03 2017 09:20
@zelig I will join the SW3 group though not sure I can attend the meeting the Thursday .will get updated.
Anton Evangelatov
@nonsense
Oct 03 2017 10:20

@zelig I’ve been thinking about the discussion yesterday during the NetworkSim demo, about being able to track and visualise individual requests for GET/POST of content and chunks.

At the moment I find it hard to get the same behaviour from Swarm on my local node and when hitting swarm-gateways.net - it would be handy if there is a visual tool or at least a trace when you request a file.

If we have something like a unique request id we could piggy-back it when retrieving a key/hash, and visualise the whole trace of a request within a cluster we control (just for test purposes). For example attaching the request id as part of log lines, together with atomic increment / step, so that we can review one request from one swarm node to another.

Is this something you might find useful, or have you discussed something similar?

@nolash ^^
Anton Evangelatov
@nonsense
Oct 03 2017 10:28
actually I see we already have generateId, however it is part of swarm/network, whereas if it is generated immediately upon receiving the origin HTTP request, tracing would be easier
Viktor Trón
@zelig
Oct 03 2017 10:39
that is a very good idea indeed
@nonsense you wanna jump into deep water?
Anton Evangelatov
@nonsense
Oct 03 2017 10:40
i can give this a try and see if i can code something meaningful
Anton Evangelatov
@nonsense
Oct 03 2017 10:46

another question i have - https://github.com/ethereum/go-ethereum/pull/15198/files - i see that you are moving away for using the global logger and introducing one that belongs to the Node. what’s the rational behind this? to me it looks like the log structure is all the same, and we might as well be using the global log, provided it is initialised immediately when starting the node. every service that needs a log from then on, can just use it, rather than embed it everywhere.

i guess you had a reason to plug it in Node though

Viktor Trón
@zelig
Oct 03 2017 10:48
look at what i pushed in pss-networktest
I added a variant of SimAdapter that uses OS socketpair based pipe for simulated connections
in memory
instead of net.Pipe
it seems after ulimit changes
btw you pam_limits stuff worked :clap: thank you
Anton Evangelatov
@nonsense
Oct 03 2017 10:50
oh cool
Viktor Trón
@zelig
Oct 03 2017 10:55
hop on the swarm hangout please?
Lewis Marshall
@lmars
Oct 03 2017 10:55
@nonsense we are passing a logger around because the network sim has multiple nodes running in the same process, so it isn't clear which node the logs come from
Viktor Trón
@zelig
Oct 03 2017 10:56
@lmars can you hop on swarm hangout too
good to catch you just a 10 min call if you busy
Lewis Marshall
@lmars
Oct 03 2017 10:56
sorry we have a JAAK call in 4 mins
Viktor Trón
@zelig
Oct 03 2017 10:56
oh my
@nonsense @lmars @nolash can we do a call at 2pm or after the pss demo only?
Anton Evangelatov
@nonsense
Oct 03 2017 10:59
fine with me
@lmars thanks for the clarification
Lewis Marshall
@lmars
Oct 03 2017 10:59
sure, so is there a schedule for these demos? I didn't realise there was one today too :)
Viktor Trón
@zelig
Oct 03 2017 11:00
i published in 4 groups twice each approxx :)
should put it on the working groups wiki
and anyway we need a calendar
we need Balint very very soon
sigh..
Lewis Marshall
@lmars
Oct 03 2017 11:03
ok sorry I completely missed them, so are they posted in here somewhere?
(or please re-post :))
Viktor Trón
@zelig
Oct 03 2017 11:05
FYI: https://github.com/ethersphere/swarm/wiki/Working-groups
1) Network simulation: monday 2nd October 3pm CEST
2) PSS demo: tuesday 3rd October 3pm CEST and planning on 5th October 4pm
3) SW3 group: thursday 5th October 5pm
all on hangout https://hangouts.google.com/hangouts/_/ethereum.org/swarm
Lewis Marshall
@lmars
Oct 03 2017 11:06
great thanks, sad I missed the one yesterday :/
Aron
@homotopycolimit
Oct 03 2017 11:39
Let us have one communal calendar where we input any and all hangouts and meetings
Aron
@homotopycolimit
Oct 03 2017 11:49
can we get a shared calendar please from the ethereum.org/swarm google account?
Viktor Trón
@zelig
Oct 03 2017 12:00
i am creating a google can from etherum
Making your calendar public will make all events visible to the world, including via Google search. Are you sure?
@lmars a quick word on the swarm hangout please?
Viktor Trón
@zelig
Oct 03 2017 12:16

parameters that are sensitive:

  • dialBanTimeout in p2p/simulations/network.go
  • kademlia retryInterval set in swarm/network/simulations/discovery/discovery_test.go

  • kademlia keepAliveInterval set in swarm/network/simulations/discovery/discovery_test.go

  • bzzHandshakeTimeout in swarm/network/protocol.go
Lewis Marshall
@lmars
Oct 03 2017 12:17
@zelig I can join now if you want?
Aron
@homotopycolimit
Oct 03 2017 12:48
@zelig public calendar? sure. why not. There is nothing sensitive in there. Besides, this channel is already public and we are constantly pasting the meetings and hangouts links here.
Viktor Trón
@zelig
Oct 03 2017 14:15
brilliant stuff thanks @nolash
Lewis Marshall
@lmars
Oct 03 2017 14:16
:clap:
lash
@nolash
Oct 03 2017 14:17
:)
Viktor Trón
@zelig
Oct 03 2017 16:17
ethereum/go-ethereum#15238 thoughts?
Anton Evangelatov
@nonsense
Oct 03 2017 16:20
bzzh sounds good to me
holisticode
@holisticode
Oct 03 2017 16:21
I second this - semantically cleaner, less ìfchecks in the API handler
Viktor Trón
@zelig
Oct 03 2017 16:42
@/all what does everyone do after devcon?
shall we stay a week hacking together?
Aron
@homotopycolimit
Oct 03 2017 16:43
I'll be travelling on to NYC on the 6th, return to Europe around Thanksgiving
Viktor Trón
@zelig
Oct 03 2017 16:45
why i ask so late this, we had thought of this long ago
@holisticode @homotopycolimit free for a call re waivers?
Anton Evangelatov
@nonsense
Oct 03 2017 16:46
i don’t have anything planned, but have a return ticket to Zurich for the 6th.
Aron
@homotopycolimit
Oct 03 2017 16:46
yes, free.
holisticode
@holisticode
Oct 03 2017 16:46
yes
Daniel A. Nagy
@nagydani
Oct 03 2017 16:48
@zelig staying in Mexico after devcon is hopeless.
Viktor Trón
@zelig
Oct 03 2017 16:48
usual
Daniel A. Nagy
@nagydani
Oct 03 2017 16:48
All accommodation is booked up because of this festival of death thing.
Viktor Trón
@zelig
Oct 03 2017 16:49
@nagydani ok i realise
Viktor Trón
@zelig
Oct 03 2017 18:07
@lmars and @nonsense have you had a chance to track down some issues re socket conn simulations?
Lewis Marshall
@lmars
Oct 03 2017 18:12
I've not had chance to look yet no, I plan to review the code later tonight
Viktor Trón
@zelig
Oct 03 2017 18:19
thanks a million man
Anton Evangelatov
@nonsense
Oct 03 2017 21:10
no success yet on my side
holisticode
@holisticode
Oct 03 2017 23:10
is there a way to use ENS resolution in a test?
Lewis Marshall
@lmars
Oct 03 2017 23:44
or just wrap ENS resolution in an interface that can be mocked in the test
holisticode
@holisticode
Oct 03 2017 23:46
:+1: thanks!