Huh? Have my peers stopped working for you @Southside ?
2 Likes
No -
I have been nuking .local/share/safe a few times cos I have been swapping between ADSL and fibre and the ISPs router and something more robust. and I deleted the txt file with your peers.
I just forgot the peers was still there further up the thread. 
2 Likes
The Node launchpad is far from Beta ready?
Not able to restart node or remove nodes, not keeping track of active nodes, option to add port range and so on.
1 Like
Correct, as long as your ISP handles a 3rd party router because a lot of ISP are using PPPoE or other protocol and routers can easily be set to use that. When ever you replace the ISP router you need to take note how the ISP router is connecting. But you say its working then ignore that since its been done (somehow)
EDIT: oh you said your router is in bridged mode, then its still doing the PPPoE for you. You maybe limited by the ISP’s own CGNAT (or whatever its using) but one expects that to be much better than any router they give you
The OS should be able to handle hundreds of nodes. If not thousands since it has main memory to use and doesn’t need any NAT table like a router does, it sends port data to the program listening to that port.
3 Likes
Exposing the computer direct to the internet isn’t that a potential security concern, without firewall and such?
Without doubt.
But if I read it correctly @Dimitar is putting machines on that have no other use. So if the firewall doesn’t stop something its only the loss of that machine till its rebuilt.
Also remember that opening ports (incl uPnP) also has risks similar if the node software can be cracked
5 Likes
They use DHCP and have attached the IP they give me to the MAC address of the lan card, is this different from PPPoE?
For the test, I requested Internet from 5 different providers, 2 local ones allow me to directly connect the cable to a computer and from there I started without a problem 41 nodes/each. Of the three other national ISPs, 2 whit optic allow Bridge mode, and the third one only offers internet through the 5G network, but I also requested it for testing.
I have turned on the firewall on all of them, with all ports prohibited except 22 and 4000, through which I access them. The strange thing is that port 5353 was added to one of them, which I didn’t add, and I wonder if Ubuntu didn’t add it by itself for some service… but on the other 2 laptops it didn’t turn on and they are identical on all settings…
Privacy. Security. Freedom
4 Likes
I would be nice if we could open a new topic on Node security, best practices and such. Example, would it be good to run nodes in a virtual machine or what can you do to add security and such.
2 Likes
Sounds strange, I would probably close that port. If the others don’t have that open then you don’t need it.
1 Like
I googled max NAT table translation entries or something like that and got.
“What is the limit of NAT in Cisco?”
“The NAT session limit is bounded by the amount of available DRAM in the router. Each NAT translation consumes about 312 bytes in DRAM.”
So 10 000 connections, 10 000 * 312 bytes = 3 120 000 bytes = 3.12 MB. If my router have like 125 Mb free memory then it could theoretically handle 125/3 = 41, 41 *10 000 = 410 000 NAT entries?
Without the listening ports of the nodes open they might not earn anything. Reason being the nodes can only contact others but not receive random requests from clients wanting to upload new chunks.
*unless they are started in home-network mode. If you used launcher then its most likely in home-network mode since no open ports allowed.
You can check by looking at the safenode-manager’s json file and you’ll see the method of connection
1 Like
Port 5353 @dimitar
It provides DNS-based service discovery for your local network. UDP port 5353 needs to be open in your firewall to allow it to answer multicast DNS
Very much depends on the software and often a set limit and remember you need storage for many other things too, like packets to be forwarded when routing
2 Likes
I wrote a question in the TP-link forum, hope to get some answers what the limit might be and limiting factors.
1 Like
Should everyone have port 5353 port forwarded if nodes not started in home-network mode?
This solved the problem for both uploads and downloads. And now I also know what confused me the most: The thing that the output of upload command can not be trusted. The successful upload did not differ from unsuccessful ones in any way:
Successful:
13 chunks were uploaded in the past but failed to verify. Will attempt to upload them again...
Splitting and uploading "ubuntu-18.04.4-desktop-amd64.iso" into 13 chunks
**************************************
* Uploaded Files *
* *
* These are not public by default. *
* Reupload with `-p` option *
* to publish the datamaps. *
**************************************
"ubuntu-18.04.4-desktop-amd64.iso" 7655f9a3a60838cfde45d6ed3bd5d916244f2aae033379384852e3b69e3efc87
Among 13 chunks, found 13 already existed in network, uploaded the leftover 0 chunks in 44 seconds
**************************************
* Payment Details *
**************************************
Made payment of NanoTokens(0) for 0 chunks
Made payment of NanoTokens(0) for royalties fees
New wallet balance: 0.999004790
Previous unsuccessful:
Files upload attempted previously, verifying 4057 chunks
1119 chunks were uploaded in the past but failed to verify. Will attempt to upload them again...
Splitting and uploading "ubuntu-18.04.4-desktop-amd64.iso" into 1119 chunks
**************************************
* Uploaded Files *
* *
* These are not public by default. *
* Reupload with `-p` option *
* to publish the datamaps. *
**************************************
"ubuntu-18.04.4-desktop-amd64.iso" 7655f9a3a60838cfde45d6ed3bd5d916244f2aae033379384852e3b69e3efc87
Among 1119 chunks, found 1114 already existed in network, uploaded the leftover 5 chunks in 3 minutes 29 seconds
**************************************
* Payment Details *
**************************************
Made payment of NanoTokens(1559) for 5 chunks
Made payment of NanoTokens(274) for royalties fees
New wallet balance: 0.999007317
Both of them appear the same. Is there something wrong with the verification and retrials, that happens during the upload? Or are they possibly turned off by default?
2 Likes
I am not sure. But since its for multicast then I doubt its needed for a node only computer
I use Termux and the Andronix to install Ubuntu server on my phone.
@Traktion reported some progress using only Termux but I didn’t getting it fully working that way.
3 Likes
Bare in mind your router needs a CPU capable of handling it all to.
1 Like
not sure, maybe just different upload status got recorded locally.
So the situation is that when I upload and the upload fails:
- I don’t know that my upload has failed.
- I only find out by downloading.
- But if download fails, I don’t know if it is because:
- Upload has failed
- Download fails because too large a batch
- Data has been lost from the network
Is there any possibilites to make this more straightforward?
By the way, I just found out that now that I have 20 nodes running, downloading a file that succeeded yesterday, fails even with batch-size 2.
Downloading with batch-size 1 worked. Took about 30 minutes for 1.2GB file.
3 Likes