• Forum has been upgraded, all links, images, etc are as they were. Please see Official Announcements for more information

Increased Network traffic on MNs since 11.2.x?

spinner

Member
Has anyone else noticed a big uptick in network traffic on their MNs? Pre 11.2.x my MN was using <15GB/month in bandwidth.

May's usage was 90GB, and already only 4 days into June I'm seeing 11GB of data transfer over the past 4 days.

This is resulting in an AWS bill of around $10/month, which is *absolutely* not a bit deal, but *relatively* that is up about a 1000% percent from the standard $0.01/month on Amazon free tier.

Any ideas? This looks to be a month-over-month increase starting around March.
 
Has anyone else noticed a big uptick in network traffic on their MNs? Pre 11.2.x my MN was using <15GB/month in bandwidth.

May's usage was 90GB, and already only 4 days into June I'm seeing 11GB of data transfer over the past 4 days.

This is resulting in an AWS bill of around $10/month, which is *absolutely* not a bit deal, but *relatively* that is up about a 1000% percent from the standard $0.01/month on Amazon free tier.

Any ideas? This looks to be a month-over-month increase starting around March.

Getting quite a lot of spamming masternode messages in debug, not sure how much of the data that is but never used to notice this error before.
 
Has anyone else noticed a big uptick in network traffic on their MNs? Pre 11.2.x my MN was using <15GB/month in bandwidth.

May's usage was 90GB, and already only 4 days into June I'm seeing 11GB of data transfer over the past 4 days.

This is resulting in an AWS bill of around $10/month, which is *absolutely* not a bit deal, but *relatively* that is up about a 1000% percent from the standard $0.01/month on Amazon free tier.

Any ideas? This looks to be a month-over-month increase starting around March.

This was already answered somewhere here on forum that it might be the result of active mixing but here is another suggestion:
If your MN is stable then most likely you have a lot of connections which is nice to be reachable and respond correctly but you could get more messages/traffic. Default number of connections is 125, some guides advise to set it even higher - 200 or so. Run "dashd getinfo" and look for "connections: ...". If it's relatively high - try to lower it: add "maxconnections=N" to your dash.conf (choose your N, I would recommend to start from half of your current number of connections) and restart daemon. Let it run for a few days until you reach that limit (dayX) then let it run a bit more and compare results starting from dayX.
 
This was already answered somewhere here on forum that it might be the result of active mixing but here is another suggestion:
If your MN is stable then most likely you have a lot of connections which is nice to be reachable and respond correctly but you could get more messages/traffic. Default number of connections is 125, some guides advise to set it even higher - 200 or so. Run "dashd getinfo" and look for "connections: ...". If it's relatively high - try to lower it: add "maxconnections=N" to your dash.conf (choose your N, I would recommend to start from half of your current number of connections) and restart daemon. Let it run for a few days until you reach that limit (dayX) then let it run a bit more and compare results starting from dayX.
connections: 127

Ill halve that and check back in a few days to see if the bandwidth has dropped. doing about 3GB/day atm, so if there is a straight arithmetic drop, that would take it to 1.5GB/day, which would still be 3-4x higher than pre 11.2.x

Not a huge deal, but interesting all the same! Will report back, thanks for the help Udjin
 
This was already answered somewhere here on forum that it might be the result of active mixing but here is another suggestion:
If your MN is stable then most likely you have a lot of connections which is nice to be reachable and respond correctly but you could get more messages/traffic. Default number of connections is 125, some guides advise to set it even higher - 200 or so. Run "dashd getinfo" and look for "connections: ...". If it's relatively high - try to lower it: add "maxconnections=N" to your dash.conf (choose your N, I would recommend to start from half of your current number of connections) and restart daemon. Let it run for a few days until you reach that limit (dayX) then let it run a bit more and compare results starting from dayX.

Does this mean some masternodes are doing quite a lot more work than others ?
(I thought it was random and balanced out)
 
Does this mean some masternodes are doing quite a lot more work than others ?
(I thought it was random and balanced out)
If you define "work" as relaying messages and providing blocks for new clients to sync then I guess the most stable MNs should do slightly more "work" just because they have more stable connections and peers prefer more stable nodes instead of unstable ones. But once again - that was a suggestion, I have no real data.
 
Since AWS is hosting the majority of MNs I would assume this is happening to a lot of other MN operators whether they are aware of it or not.

Udjin, is there a place to check out known issues to see if there is a definitive answer? I'd be interested to know the difference pre/post 11.2 that was causing such a big increase in MN traffic.
 
Default number of connections is 125, some guides advise to set it even higher - 200 or so. Run "dashd getinfo" and look for "connections: ...". If it's relatively high - try to lower it: add "maxconnections=N" to your dash.conf (choose your N, I would recommend to start from half of your current number of connections) and restart daemon.

I've reduced maxconnections down from my own default value (=256) for 2 of my Masternodes and report back once the numbers have stabilised.
 
Since AWS is hosting the majority of MNs I would assume this is happening to a lot of other MN operators whether they are aware of it or not.

Udjin, is there a place to check out known issues to see if there is a definitive answer? I'd be interested to know the difference pre/post 11.2 that was causing such a big increase in MN traffic.

I doubt there is any place you can find it. It was reported in one of threads here on forum and/or on btct few times I believe.

Earlier we used to use Jira http://jira.darkcoin.qa/secure/Dashboard.jspa to keep all issues in one place but now it's pretty much abandoned (though there are efforts to bring it back to life). Alternatively you can submit new issue on github https://github.com/dashpay/dash/issues (some guys here gonna hate me for this :tongue: but I personally prefer that way, because 1) I think issues should be kept where code is and 2) though Jira is way more powerful for managing issues, that power comes together with complexity imo).
 
Looks like dropping the maxconnections had a big effect - currently my MN is sitting at 13 connections and the bandwidth usage has taken a nosedive. crowning, are you seeing the same?

Screen Shot 2015-06-05 at 3.02.35 PM.png
 
Looks like dropping the maxconnections had a big effect - currently my MN is sitting at 13 connections and the bandwidth usage has taken a nosedive. crowning, are you seeing the same?

View attachment 1418

Looks so, but you need to wait some more days because it needs some time until the connection number hits the maximum.
I'll wait until my nodes hit my current maxconnections before I draw conclusions.
 
Update - Connections up to 60 and network traffic steadily increasing.

Can anyone explain whats the importance of having a high # of connections? If the bandwidth consumption of having a high number of connections is actually affecting ROI of the MN, whats the side-effect of dropping the max# connections to a very low number?
MNweektx.JPG
 
Update - Connections up to 60 and network traffic steadily increasing.

Can anyone explain whats the importance of having a high # of connections? If the bandwidth consumption of having a high number of connections is actually affecting ROI of the MN, whats the side-effect of dropping the max# connections to a very low number?
View attachment 1438
There is an idea (myth?) that more connections i.e. better connectivity gives network lower block, tx etc propagation times. That might be true for relatively low number of connections but it was never proved that further increase gives some reasonable improvement afaik. I personally think that 8 (current max number of outgoing connections) is ok and say 16 connections in total is already enough. But that's just a feeling, I have no proof for that too :wink:
 
There is an idea (myth?) that more connections i.e. better connectivity gives network lower block, tx etc propagation times. That might be true for relatively low number of connections but it was never proved that further increase gives some reasonable improvement afaik. I personally think that 8 (current max number of outgoing connections) is ok and say 16 connections in total is already enough. But that's just a feeling, I have no proof for that too :wink:

My Masternodes normally settle at around 90 connections each.
 
The Bandwidth was creeping up up up on my MN so I set the max connections at 16.

I'll report back here if I notice any weirdness over the coming weeks.

crowning did you see the same behavior with your bandwidth consumption?
 
The Bandwidth was creeping up up up on my MN so I set the max connections at 16.

I'll report back here if I notice any weirdness over the coming weeks.

crowning did you see the same behavior with your bandwidth consumption?


The traffic is (roughly) a function of the number of network connections, and in theory limiting this number will decrease traffic...BUT, I noticed that my testnodes happily ignore the maxconnection setting, I've set it to 32 and they are up to 44 and 50 again after 2 weeks.

I'll wait how v0.12.0.x behaves, and if it's the same I'll investigate why the numbers go beyond the set limit.
 
The traffic is (roughly) a function of the number of network connections, and in theory limiting this number will decrease traffic...BUT, I noticed that my testnodes happily ignore the maxconnection setting, I've set it to 32 and they are up to 44 and 50 again after 2 weeks.

I'll wait how v0.12.0.x behaves, and if it's the same I'll investigate why the numbers go beyond the set limit.
I'm seeing the same thing on my side - I set maxconnections = 16 and I'm already back to almost 40.

My free year with AWS is up this month anyway so all of this is making me think of pulling the MN back private.
 
Back
Top