Question: What is the best way to monitor the total number of nodes? I suspect the table only shows those that are in my group.
One method might be to have a script that pings all known nodes to determine if they’re still alive, and periodically restarts the vault in order to collect a new group, then pipe just the total number of nodes to a graphing program for public viewing. I’ll see what I can do.
Thing is, for a production network we don’t want anyone creating a list of all known nodes, or does that only matter while the network is small. If there were 10^8 separate IPs then let them choke on such a list.
I’m not sure total nodes can be known in an efficient way.
periodically restarts the vault in order to collect a new group
I doubt that would work well… but if you can see the node.log, perhaps you could script a sum of the number of nodes it interacts with over time in some useful way. Pinging them is not an ideal approach… and I don’t know the value of knowing how many of the nodes you’ve seen are still live. Quicker would be to request the same data from other users, as geography will see nodes faster than you will. Plotting ip addresses with usual tools might be useful indicator of uptake in various regions of the world.
I expect there are more interesting and useful data to resolve from what can be known by the local node. For example, how much of the local node and it’s neighbours resources are being used?.. If the network spreads the load, that might be an indication of demand cf capacity.
What other stable metrics might be available, both locally and global, to indicate network performance; load; and stresses?? Can we sum detail from various locations in a useful way for that?
How would we know if the network was stressed?.. user experience is not a good responsive metric.
I do wonder that some clever so’n’so could recraft a node to do more than default and that could be pinging data about it’s experience which alongside others might suggest lag and load etc.
question: is it possible to add an authentication screen before the game/script starts, so that one would have to authenticate with the launcher to play?
Seems like a good way to try my hand at api stuff.
Very awesome that this is happening. These efforts are very important, and it’s great to hear that the next MaidSafe release will allow many testnets to be run by us all with relative ease
Constantly-available community-nets are extremely important
Yes, it is, but I don’t know how. Check the code on the decorum project. That would be a good place to start. PLEASE NOTE HOWEVER: how data is handled on the network is going to change dramatically from what is currently used there. I’d not use it as a best practice site, merely to see how people are interacting with the launcher.
@bluebird I would guess that because MaidSafe controlled all the “hard coded contact” nodes, they used that as a metric and counted how many individual vaults/IP connected. Maybe counted distinct IPs? I ran a few vaults, so I hope I wasn’t counted 3-4 times.
Whoa whoa whoa! www.wes.safenet is back! Are archive vaults a thing already? It wasn’t showing up because of “get failure” so I assumed it was gone forever (non-persistant vaults) But… its back. Confused, but very pleased with the network resilience.
Looks a little bit ahead of what I currently understand. I can conceptualize what I see there, but not really know where to put it. Haha.
Learn as long as we live aye?
I put it up the same time that www.hello.safenet went up two days ago. We both lost our sites within an hour or two of each other (can look up this thread). Not long after that, I lost access to the account tied to that site. Now the site is back, but still no access to the account tied to it. Interesting.
Interesting. I wonder if its because vaults were at their limits so the network sacrificed some data or actual data loss from churn / too many vaults going offline at one time. I hope its the latter. If its the former, I’d like account data to be last on the “delete me” list.
It is only taking 5-8 percent of my cpu. That’s impressive. It is not memory intensive as well. It is taking .5 percent to 5 percent.
Edited:
Every time data is being added to the vault, the memory spikes to 5 percent. Once the data is done being transferred to the vault, it heads back down to .5 percent.
What are the specs on your machine? I have 10 vaults running with a boatload of browser windows, messing with demo app etc and only use about 5% of the CPU on average (total, not vaults). I haven’t seen a vault take more than 1% at any given time.
IMO, for what the end goal is (I know, we’re testing, still alpha, just saying) using 5% of the CPU at all times is no bueno.
You should not need to set up any DNS servers. Once the launcher is up and running on your computer, the network handles all the “DNS” for you. If you use the IP’s in the original post you should be fine to test it out. Just remember this is all alpha level software.
