Opinion on current traffic?

How to check if there are other nodes in your /24 subnet:

Start a second node. If there are no other nodes, then the combined traffic of both nodes will be about the same as with just one node.
If there is one other node, then the combined traffic will be greater (by about 33%) than with just one node.

How this would work:
Let’s say that each /24 subnet gets 10mbps ingress.
If you run one node and there are no other nodes in your subnet, you will get that 10mbps.
If you start a second node, each node will get 5mbps, for a combined ingress of 10mbps.

However, if there is another node on your subnet, your single node will get 5mbps (half of normal).
If you start a second node, now there will be three nodes on your subnet, each getting ~3.3mbps, since you own two of them, your combined ingress would be 6.6mbps.

only works if that 2nd node is already vetted.


the numbers will only be reliable for a subnet if both repair and normal ingress is used.
else they will seem almost random…

i would be very happy if i don’t share a subnet with somebody else, because then i don’t need to deal with it as a high priority.

might try @Pentium100 test see what that does mainly because i’m making a new node anyways when the stars align… got my nice new pcie ssd… but ofc the drivers for the version of debian i use are homebrew so doesn’t seem to work… which kinda sucks… lol

maybe the lower ingress is a geographical thing, just seems odd now that the bandwidth comparison thread was running for so long and the numbers where basically the same all the way through…
that it should be different within a couple of releases… but maybe storj changed something like was said in the townhall.

but at this pace my node is basically stalled in growth, ofc i shouldn’t complain because i got fairly decent egress :smiley:

i even tried to under allocate space for the node… to check that it wasn’t because my max capacity was a bit over what was available on the array / pool

You can try to do a scan of all the other IPs on your subnet on the standard storj port to see if there’s other people. Many more variables to ingress as well. If there is just one other node I don’t see why they would have a non default port but if they have more than one then it’s harder to tell exactly how many there are.

Here’s my traffic graph.

At least egress dropped before the first outage, so I know it’s not because of the downtime.

Fun fact. As there are around (256 - 32 - 3) * 256 * 256 = 14483456 blocks of /24 subnets, due to the birthday paradox, random block C collisions start to happen with non-trivial probability already at around 2000 blocks with nodes. We’ve likely crossed this threshold already.

Actually, likely far earlier, as probably far less block C addresses are actually available to potential SNOs and fulfil Storj requirements for speed/bandwidth.

a 5 or so hours into the new day statistics … would be nice if it was easy to see what time storj is on but i’m guessing GMT so either +1 or +2 and its past 7 am.

so atleast 5hr and 20 min into the new day and my ingress is still in MEGABYTES!!!

if it wasn’t so sad it would be funny, so yeah it dropped down even further

or more of a @general discussion… i have been seeing this in my logs for about as long as i have gotten half ingress

DEBUG orders sending
2020-09-12T05:41:05.242Z DEBUG orders no orders to send
2020-09-12T05:41:05.242Z DEBUG orders no orders to send

not sure what that means, but it’s been very consistent and i’ve not noticed it before recently.
does something else there.

i read some stuff about that i think like 30% of C blocks are allocated to special usages or super large internet users, its only in recent time that people would have to pay for leasing them, in the past companies like say microsoft could just say, we need 1mil ip’s and most likely not even need to give more reasons than we are microsoft, and these blocks are still allocated to this day, even tho some major companies have given up their blocks to help ipv4 survive a bit longer.

@NNick yeah thats a great idea, usually scanning the internets is a bit of a monumental task so my mind kinda just ruled it out…

and yeah i got 1 ping back from something located on a port 28967 in my subnet…
i wonder what that is … i would say if you are seeing the same about of ingress as i am, then you must also be sharing your subnet with another ip / node.
how data is distributed between those i have no data on… so from what you are saying, then the distribution between nodes on a single ip, only happens after the ip is chosen…

or it has to, or that would make it so that data would be evenly distributed between the different ip addresses in a subnet, thus explaining why you changing the number of your nodes doesn’t change the ingress ratio.

and basically sinking @Pentium100 idea of how to check for nodes sharing a subnet

but i do understand you brightsilence, i wouldn’t have expected to other nodes on the subnet for a long time… but i guess reality had other plans…

on the positive side we learned something… it seems that the rule is that each ip gets its share of a subnet’s ingress… which makes sense, if one is building the system, one would want even distribution between different parties on the network, and that seems like the obvious way to ensure or atleast aid that goal.

The birthday paradox doesn’t apply here. We’re not looking at the chance of a collision happening to anyone in the group, but to a specific individual. The chance of that is about 1 in 2 million. Of course that assumes there is an even distribution of SNOs among IP subnets. It also assumes that all about 8000 nodes are on their own subnet, which we know isn’t true. Se the odds are probably not that small.

This is normal. Since orders were moved to files I believe it tries to send orders every 5 minutes but there is only something to send once every hour. This was mentioned in the change log.

Now that is quite a bit stronger evidence that you may have been very unlucky. Perhaps not coincidence. Have you encouraged people nearby to start running storagenodes?

Code says no. It picks a subnet and then a random node within the subnet. So @Pentium100’s suggested test is valid.
I was only counting normal ingress earlier. So I have to retract what I said about seeing the same numbers as you did.

That’s the thing about jumping to conclusions. You’re assuming all data you base it on and the final conclusion is true. But in this case you were simply working with bad data. (My bad in this case) Luckily code is absolute and I can say with certainty that the code only deals with subnets and not actual IPs during node selection. So this conclusion is absolutely untrue.

Likewise when the chances of a collision are 1 in 2 million, you really need strong evidence before accepting a conclusion that there is another node in your subnet. Now an IP responding to port 28967 is much stronger evidence than you presented before. So yes, I am now inclined to believe there might be another node. Still less inclined to believe that was entirely coincidental. I’ve personally taken the risk of encouraging people nearby to start a node. You may have done the same and now have shot yourself in the foot in the end.

well if it cannot work that way then your numbers cannot be correct…
both cannot be true… you cannot get half ingress and have setup multiple new nodes in the last 3 months and still have the same half ingress unless if thats how it works…

but no matter, no i have not gotten anyone to run storj, but like i said at first, we are a small country and all fiber is basically the same company, so that may greatly increase my odds… it is ofc unfortunate, but maybe it’s just a temporary visitor on the subnet…

i don’t have multiple nodes yet, been trying to set it up, but my containers cannot run docker because the docker on the host affects it in some way… :slight_smile: one of those impossible things according to some, that i seem to be averaging 3 of pr month.

i think it’s the internet here… we don’t have that many isp’s and most of them share infrastructure to a large degree, and also we have super high bandwidths basically nation wide you can get fiber, so people will ofc be interested in projects such as these… on top of that the last townhall basically just launched the ip war since we are now allowed to use all the ip’s we want

You mean the statement I specifically retracted in my previous post?
I was counting only normal ingress which has been roughly half of total ingress this month.

I’m talking about my perspective. From my perspective it’s an approximation of the chance someone will start complaining about low ingress on the forum. Might be pretty accurate ^^

i will admit the odd’s doesn’t seem very good, but it’s very relative… like say how big a ip block does my isp actually have… and still it’s shared with like a couple of hundred devices and the odds one of them is a storj node seems pretty low…

ofc that math changes a lot if we start thinking about some of the other variables… like say… how many users at the isp actually request a static ip address… and then are all those taken from the same pool which may be fairly limited… when i scanned the network i could see computers within 50km radius…

so somebody with the same isp as me, requests an ip addess and because we are both with in a certain time frame we get bundled up in the same subnet, and thus the odds of over lapping subnets all of a sudden might become much more plausible.

a simple solution would be for me to try to switch to a different subnet, and get hope it doesn’t happen again…

so seen from a bit more local realistic perspective, there can be 250 device or so in a subnet and if really the odds would be calculated on how big the odds are that one of those would attempt to run a storagenode.

all the other factors are in a sense irrelevant to that ratio, because its mostly static who or what will be on the subnet… if the ip’s are assigned in a sequential manner, then if storj is popular in one period in time like it has been recently, then more will adopt and end up in the same subnets because of how the isp maybe distributes the addresses…

but yeah atleast for now there should be plenty of room… the SNOboard could use a shared subnet indicator tho… took me a month to notice… many might not even find out and just assume their node is slow or whatever…
and the only reason i could see it was because i made that bandwidth comparison thread…

atleast it’s been a slow month for ingress, so haven’t missed much :smiley: and egress have been fairly nice.

How did you quantify this? Namely distance vs IP address since that relies on mostly “best guess” geo-location data provided by ISP or other 3rd party.

Between Bright’s points and yours, it seems highly likely of an organic addition from being on a fairly, your, nationwide ISP that is basically the de-facto choice in a nation small enough to have two SNO’s easily operating within the same class C subnet.

they used initials of their city in their dns name… but yeah else i wouldn’t have any idea… and might not be correct… but there was a few of these cases which gave me a idea of the reach of the network i am on… else i could have traced the ip, which is also usually pretty good for atleast seeing some parts of where stuff is going so long as it’s plotted on a map, also it’s inside my local county mostly and the company is a regional company operating until recently within less than a 100km radius.

not sure i really understand what you are saying here, but from how i see it the whole subnet odds
comes down to what the odds are that one of the 250 odd other computers on the subnet is going to start up a storjagenode… which again could be largely affected by how the ip’s are allocated, how the local infrastucture looks and how much spare tech people has, local internet bandwidth, and ofc last but not least, storj popularity.

these factors if aligned correctly might make it no problem at all… and it also might make it a problem almost every damn time… i mean right now… if somebody with a 100km radius calls the isp i got and requests and ip address for a storj node… then i would almost be certain they would pop up right in the subnet that i currently share… and since i’ve had this for 7months and if its counting down in ip range then it might have 7 months left until the subnet is filled…
after that the odds will become immensely lower… but because i’m in a subnet thats still being filled by people requesting static ip’s…

then the odds might be pretty good or bad depending on how one looks at it…