There’s only one topic of conversation this week, so the update is a bit shorter and less detailed than usual, although as you can imagine, it’s all go in the engine room.
So, for sub-surface dwellers, the big news is that the Beta Network has launched and the first wave of testers are up and running. It’s taken a wee while (slight understatement), but we got there, just as David always knew we would. And when we say we, that means all of us. It could never have happened without the steadfast support of the community over the years.
Thanks to you the network has grown from our original 2,000 MaidSafe nodes to over 17,000 nodes and counting (more seem to be attempting to join!), our biggest ever by some margin. The first wave of testers providing this impressive sea of storage are mostly patiently awaiting their nanos. Inevitably there are still one or two glitches to iron out, but that’s one of the reasons to have a phased Beta testing program — we figure the first wave will likely be the most technically advanced.
One of those glitches is the laggy DAG, which is slow in delivering the nanos to the Discord Bot
. In part this is because of the rapid expansion of the network, and we have a fix in play to address this. Rest assured, though, that all your nanos are being logged and stored on the DAG. And we have performance improvements waiting in the wings. We just need to tie it in with the Discord Bot
without breaking things.
Another is the overload of the bootstrap servers. We normally launch 2,000 nodes across 50 machines, and that’s never approached 100% CPU before. Here we see that the bootstrap nodes alone would be hitting 100% CPU, and this is undoubtedly causing issues for folks attempting to join and for uploading data. We’re looking to have bootstrap only nodes in the contact file, on dedicated machines now. And then to have other “capacity” nodes when needed. (We may also look to diversify the contacts file, if anyone was happy to have their node’s peers shared for bootstrapping, this could help lighten the load!)
The scale of the network means the data on it is spread fairly thin. This should rectify itself in time as uploads catch up.
Another issue a few of you have reported is with nodes stuck in “added” state when using the launchpad.
Nic and @roland are looking at that now. For those comfortable with using the command line, the safenode-manager is another way to launch nodes which doesn’t seem to have this problem. Let us know if you need a walkthrough for this.
Some have experienced an error using the safenode-manager
and some Windows users are finding nodes are not running. Again we are looking into these issues.
We’re building in more monitoring in all areas of the code, so finding these bugs and bottlenecks is getting easier. The Node Launchpad
is next in line.
Massive thanks again to all you first wave testers. Keep the reports coming. It’s all valuable info and the team and community members are happy to help and to answer any questions if you get stuck. If you didn’t make it into the first wave, make sure you sign up for the second one, when hopefully most of the current issues will have been resolved.