@qi_ma is working on a cheeky fix to be more tolerant of odd connectivity issues.
With those changes he got a 37mb “safenode.exe” uploaded to b0bd3fcef1afbf917a9b8b2936c13c9f66f6243d542643cb4353644572b5e8bc
Would be good to see if folk can get that?
any instructions on how to run on windows? what do i do after i have downloaded the exe from github?
Yes, CRC32 A84010B8.
But it is completely equal to file downloaded yesterday. So something went wrong.
Are problems happen for every node or only for some of them?
Should users restart their nodes with new binary?
Yes I could and it took only a few seconds 
./safe --peer=/ip4/178.128.166.245/udp/36572/quic-v1/p2p/12D3KooWPuaKwn7rFNn4oj1ABDpikVZ7scHs29ZNCHBveLHgCmak files download -- "safenode.exe" b0bd3fcef1afbf917a9b8b2936c13c9f66f6243d542643cb4353644572b5e8bc
Starting logging to directory: "/var/folders/z_/n_9kwvnn5f11q35z7m18c8sm0000gn/T/safe-client.log"
Instantiating a SAFE client...
Downloading file "safenode.exe" with address b0bd3fcef1afbf917a9b8b2936c13c9f66f6243d542643cb4353644572b5e8bc
Successfully got file safenode.exe!
The change is only with the client upload part.
Nothing related to the network.
No need to restart any node.
same here =) (just copied your command 
)
sha256sum: 007dd7b270922694b798b3c4f529620c8206255ca5bf91a02d1e5ca38ad1938c
(identical checksum as the released windows safenode.exe from github after unzipping it)
This may explain why files are equal.
Probably safe.exe was supposed to be uploaded instead of safenode.exe.
That’s it! The file was safe.jpeg and I tried to upload it as safe.jpg. @joshuef, sorry for the false alarm.
Would be good if that produced a specific ‘file not found’ error.
Away from my laptop ATM, but it’s great to see the test net living so long! Amazing work everyone! 
Few questions about Joshnet:
It seems to be quite stable, the structure itself I mean. And when I put my laptop to a sleep for an hour or so in the morning, and then re-opened it, the node just sprung back to life and seemed to continue as if nothing happened. I didn’t check the logs or inspect anything though, just checked from the task manager, that CPU and memory consumption indicated normal activity.
That disk write total even went down to 302 MB or something, so I don’t know what that measures…
Self encrypted as before.
Maidsafe started 25. I upped that to 35 this morning. I’m not sure of a reliable way to check that from our logs as yet. I know libp2p has some crates that offer metrics etc, we’ll be checking that out soon.
The extra 10 from maidsafe at least. Leaving, i suspect so, but cannot say.
We can kill 35 nodes, which sounds like the vast majority of the network. It would not necesarily die if the others new enough other nodes… I’d wager it would die effectively though.
With enough community nodes it would live on though. You could reup the “network” from one node. But you’d likely have massive data loss… Soo YMMV?
Yup. 8.
I asked, because I remember sometime before you could see pieces of images, or something like that. Like chunked, but not encrypted?
Very impressive, compared to how it used to be.
Who cares… 
But seriously, if we now lose say 3/8 replicants because of mass loss of nodes, does the network try to recreate more replicants if new nodes join?
Yes it should handle that.
Do you still need to port forward manually to join this?
Windows here, but struggling.
There is no NAT traversal, so public ip only for nodes I am afraid. Client’s don’t matter so apps/cli all ok
Update:
54.2 GB Received 67.2 Sent
Still running well. Just amazing!
How can numbers be so high, when node is only 1GB?
I’m not sure. Churn?
These numbers come from the Ubuntu System Monitor. Nothing else is running on the machine.
I just added 25 DO nodes… aint letting this one die 
If it is still running this evening, I will add more.
Not often you get a test in your name, 5 toes is a small price to pay!