You mean that it would be worthwhile to test the current setup even if there would be this intermittent hanging of CLI commands? That would be nice, just to get something to play with!
Adults not being reachable I guess could be leading to some of these hang issues?
The command doesn’t get executed as it can’t find the adults?
Or am I reading into this wrong.?
If there is an issue with monitoring whether certain work was done or the node handling it dropped, I wonder if a bundle of tasks, would reduce the volume of checking… send to an adult and get one ping reply back that it’s still up and completed that batch.
Good call to postpone the testnet, and you too have a good have a weekend! It is a fair weather in Helsinki, I hope that’s so wherever you are, too!
One thing I would still ask, is if you can let us know if the v6 will allow non-IGD nodes or not? If not, I could familiarize myself with my router’s IGD settings over the weekend.
Unfortunately not. People with no igd or direct connected machines will require to set port forwarding on their router to run nodes, not clients tho, they don’t need to be connectable. I hope we can find a decent howto for everyone, but it’s nightmare territory to support all routers.
What we can do is make it clear how to set the ip and port addresses in the node config to allow this. Then all folk need do is set up port forwarding.
Great resource for port forwarding and helping setup your equipment - https://portforward.com/ Used this a lot back in the day when I was in tech support and I don’t miss getting yelled at by someone sitting on a yacht complaining about their cob job Internet not working!
Curious if batching will be done with zk-rollups or in some other manner?
In the outputs, we did a neat trick. All outputs are a single sig now. So we could probably issue 100,000 transactions per second this way, but it’s cheating to claim that. However for micro-transactions knowing you can make this amount per person per second is not to bad. So how many of these can a mint create? We will soon see, but out 100input test is about 10% slower so possibly 1,900 people per second doing say 1000 transactions each, maybe more, a limit will be handling the number of clients more than transactions.
As I say let’s pull together a matrix of transaction speed in various forms and then we can get a much better picture. Lot’s of projects really screw with numbers but here we have pretty immense tps, so let’s see the worst case which is likely every transaction being 1->2 and then a bunch of 1->1,000 cases. Then we are all better informed.
BTW we played with @mav s threshold crypto which takes threshold_crypto and uses a faster sign/verify and the speedup was considerable. So we are nowhere near done yet.
Would this affect a local network?
I experienced random hanging issues on uploading data using testnet 5 binaries for baby-fleming that would go away with a restart.
I did check my router briefly before v1, and turning IGD on seemed to be almost one-click-switch, if I remember right.
What I don’t understand is if it has some implications to my home LAN in general? Like do I need to re-setup the network connection of every device for example? Is IGD a (severe) security risk in some ways?
By the way, I think my 2 years old daughter agrees with this PR
It is important to say that some people will not be able to host node.
No matter what settings they will change in their equipment.
Well done team on fixing the hang issue - on a Friday too really really looking forward to testnet v6 now. I’ve a feeling this will be eye opening
Mini update - we’ve reproduced a “hanging” issue again, so the “fix” on Friday may have resolved something, but there’s still something else.
Yesterday we also merged a fix here for a potential cache dead-lock/hang which we identified during our investigations, so it seems like there were/are multiple bugs which could lead to hanging.
The team are hard at work digging for the root cause(s) of the hangs. We’ll do an official update tomorrow with the latest.
Yes @Josh, the hanging issue’s we’ve seen are reproducible locally, though intermittent so you would not always see them, and restarting could appear to resolve (because they’re intermittent).
These hangs have been the biggest hurdle towards a stable network so far. Once these are resolved I think we’ll be off to the races with stable long living test nets. But these bugs seem to be nasty imbedded little shites. Put on your stomping boots @maidsafe
If hangs go away with restart of CLI, then it is not a big problem.
Largest problems are node crashes and data loss during splits.
And, of course, hangs inside nodes.
Think the AE work will address data loss but still needs public testing for sure.
Could we get mini updates in the updates thread too please?
Sure, I’ll try to stick to that topic
Will see what investigations throw up today and post an update summary later.
What’s the odds on v6 happening this week? My money is on less than 40%. But I’ve lost a lot of money on hunches
Good things come to those that wait, that’s what Guinness said.