Is VPN necessary? Only hoster of 88.198.210.245 knows my name. And this is plain cheap VPS.
VPN is not really necessary, just another level of indirection for the extra paranoid. If your VPN provider and VPS host are even in different countries, you need an adversary of the rank of NSA to force both of them to cooperate, or to track you down by monitoring global internet traffic. |
|
|
|
Running Nxt over torI tried to run Nxt over tor using torsocks, but apparently this doesn't work for java. The way to tell java to use a socks proxy is to define socksProxyHost and socksProxyPort properties, in a system properties file or on the command line: java -DsocksProxyHost=127.0.0.1 -DsocksProxyPort=9050 -Xmx1024M -jar start.jar
Replace 127.0.0.1 and 9050 with your tor host and socks port, if not default. Basically this is equivalent to the socks proxy options in the configuration of bitcoin-qt, if you use that. You cannot use tor if you want to advertise a public IP or run a bootstrapping node, so only use it on your mining node, if you care about preserving your privacy. And you cannot hallmark a node ran behind tor, so if you are a stakeholder and want to contribute to the network stability, please consider running a separate publicly visible hallmarked node (and if you want to anonymize that one too, need to use a VPN which allows port forwarding and dynamic DNS).
|
|
|
|
Last night fluke.airdns.org didn't crash, so I hope the tcp tuning I did yesterday fixed it. There are almost no close_wait connections at the moment. This is what my /etc/sysctl.conf looks like now: net.ipv4.conf.default.rp_filter=1
net.ipv4.conf.all.rp_filter=1
net.ipv4.tcp_syncookies=1
net.ipv4.conf.all.accept_redirects = 0
net.ipv4.conf.all.send_redirects = 0
net.ipv4.conf.all.accept_source_route = 0
net.core.wmem_max=12582912
net.core.rmem_max=12582912
net.ipv4.tcp_rmem= 10240 87380 12582912
net.ipv4.tcp_wmem= 10240 87380 12582912
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1
net.ipv4.tcp_fin_timeout = 7
net.ipv4.tcp_max_syn_backlog = 16384
net.ipv4.tcp_synack_retries = 1
net.ipv4.tcp_tw_reuse = 1
net.ipv4.tcp_keepalive_intvl = 30
net.ipv4.tcp_keepalive_probes = 5
and I also have the open files limits increased to 400000 in /etc/security/limits.conf. I noticed one weird thing, seems like blacklisted peers just accumulate without ever being removed from the blacklist. I have other machines running Nxt, and their IPs were in the blacklist of fluke, even though they were up and running. This morning everything under active peers were zombies with 0 traffic. I removed peers.nxt and restarted, to clear the blacklist. And, on a machine without public IP address, all I see in active peers are the 3-4 well-know nodes. On another one that advertises its public IP, I do see nodes other than the well-known in the active peers list, but not that many. Is all that normal? Finally, please don't put both fluke.airdns.org and nxt.airdns.org in your well-known nodes list - those two point to the same IP, I only run one bootstrapping node, not two. I intended nxt.airdns.org to be used, but somehow another DDNS name I used before, fluke, leaked out and became more well-known. |
|
|
|
Yes, you have to url-encode at least # and &, that I can think of. You can't even fix it on the server side because in the case of # I don't think the string after the hash gets submitted to the server at all. So, people with special characters in the password will have to encode them manually, see e.g. table here: http://www.w3schools.com/tags/ref_urlencode.aspOr do a simple javascript page that does the encoding in javascript on the client. Obviously, don't submit your passwords to any online services that offer url-encoding. |
|
|
|
allowedUserHosts is not enabled yet. I'll add logging feature in 0.3.16. TIME_WAIT is normal behavior, what is the timeout btw?
$ cat /proc/sys/net/ipv4/tcp_fin_timeout
5
I reduced it to 5, was 60 before the last crash. |
|
|
|
Do u have a log of requests? Such a problem will arise if someone emulates web client responses. I see 2 ways to fix the issue - add timeout for pending interface requests & enable allowedUserHosts.
I set allowedUserHosts to 127.0.0.1, but I can still access http://fluke.airdns.org:7874/ , and of course I am not connecting from localhost. http://fluke.airdns.org:7874/nxt?requestType=getMyInfo returns my correct IP. Is that how it is supposed to work? [2013-12-07 18:42:42.330] "allowedUserHosts" set to "127.0.0.1"
[2013-12-07 18:42:42.330] "allowedBotHosts" set to "*"
After a reboot, netstat doesn't show any CLOSE_WAIT connections, but about 2000 in TIME_WAIT at the moment. I don't have a log of the http requests, are those being logged somewhere? I only have whatever is printed to nohup.out, I run it from nohup. Next time it accumulates CLOSE_WAIT junk, I can capture the list of IPs from netstat. |
|
|
|
|
fluke.airdns.org ran out of file descriptors again. I bet we have a connection leak somewhere in the code, connections not being closed. There are godzillion of connections in CLOSE_WAIT state. I don't see a way to force those to timeout earlier by tweaking net.ipv4.tcp* settings, all I was able to quickly read on the topic suggests it is our application responsibility to take care of closing those. Anyone with more networking experience please feel free to suggest a fix.
|
|
|
|
Marked node doesn't need an unlocked account. But a node with unlocked account doesn't get a hallmark automatically (in current implementation).
OK, this is good. So I can unlock and mine on a node that is not public, and use a separate public node with a hallmark. Even if the public node (which presumably attracts a lot of unwanted attention) gets hacked, my private key has never been there, even only in memory, so my account is safe. |
|
|
|
|
Is there an API to verify whether a node has been marked, and with what account? Or to retrieve the node hallmark?
The node on which an account is unlocked (in order to mine) doesn't need to be marked, and a marked node doesn't need to have the account unlocked, correct?
If one marks a node, which is not behind a vpn, an attacker can correlate the marking account with the node IP, thus compromising your anonymity, correct?
|
|
|
|
yes, have raised ulimit too, running with 300+ peers fine now.
do you have any additional/usefull port restrictions to protect your node?
Not really, it is behind a vpn which effectively acts as a firewall, the only ports forwarded are 7874 and 7875. And there is nothing but Nxt running on it, with -Xmx3584M. |
|
|
|
Yes, and seems like our networking is too dependent on the bootstrap nodes. Last night fluke.airdns.org crashed with the java process exceeding the 1024 number of open files limit. At that point it had more than 400 active peers. I increased the open files limit to 16384 and restarted, this morning it had crashed again. Right now it is at 200 active peers, will see how long it lasts. |
|
|
|
|
Disappearing balance bug.
I had my account unlocked, on localhost, and mining, for some time. Suddenly, the balance shown in the top right corner went to 0.00, and the "send money" arrow got deactivated. I freaked out, assuming somebody has stolen my Nxt. Checked the transactions list, nothing has left my account. Refreshed the page, F5, logged in again - still 0 balance, yet no outgoing transactions. Only after restarting the server (without deleting any *.nxt files) and logging in again my balance showed back to normal. After a few hours, the same thing happened again. Restarting the java process fixes it, but it was quite an unnerving experience the first time.
Curiously, while my balance was showing as 0, and sending money was disabled, I was still seeing the "You can generate the next block..." message. And I even generated a block, while at 0 balance. So it may be some UI bug, the server still aware of the real account balance? I found no exceptions in the log near the time this happened.
|
|
|
|
Compare the top accounts with top miners, you will find most of the top accounts are not holding coins sleeping, but they are mining for profit to get more coins.
Don't forget that they are also mining to secure the network. The number of transactions is still very low, so for somebody already having millions of Nxt the profit from mining is insignificant relative to his holdings. They should giveaway some coins to the new comers, for example, if each of them take 10,000,000 Nxt for giveaway to 10000 users with each one 1000 coins.
This should cause the value of Nxt to plummet. Why buy something that can be obtained for free? And 1000 coins is too small of an amount to be useful for mining, so how exactly will that help those 10000 users? Most likely they will just forget about it after the novelty wears out. On the other hand, if they paid $10 for those 1000 Nxt (the current market value), they are more likely to get involved and use it. If the top stakeholders are not selling, this is because they find the current price too low. It is a free market after all. If one is serious about investing in Nxt, one has to buy it. |
|
|
|
Which port needs to be open to allow peer connections while behind a firewall, still 7874? Now that the web interface is at 7874, is there a way to block outside users from unlocking their accounts on a node that I run (because it is not secure and I don't want to encourage it), without breaking p2p connectivity? |
|
|
|
However, when I did some research on it I discovered the following things: The currency website www.nextcoin.org and NXT echange platform DGEX.com are both hosted by the same company that appears have produced the currency. That's Graviton Capital Inc. Huh? "Appears to have produced the currency"?? Did you, maybe, in your research, ever stumble upon the original thread where Nxt was announced and discussed a few months ago: https://bitcointalk.org/index.php?topic=303898.0Nxt was not produced by DGEX. Not any more than Bitcoin was produced by MtGox. And there is no official currency website (yet), the original developer didn't see the need to have one - refer to the above thread again. DGEX stepped in and created this website to promote the currency, and by this get some business their way of course, but that's it. They did not create Nxt and are not controlling it in any way. Seems like you are just trying to spread FUD in order to bring down the price and load up on Nxt while it is cheap. |
|
|
|
Nextcoin added!
Thank you, but Nextcoin is not the official name of the coin. It is just Nxt. Could you please correct that? Please consult the original post where it was announced, the creator BCNext never intended it to be called nextcoin: https://bitcointalk.org/index.php?topic=303898.0and the poll where it was voted that the name shout be Nxt (not Next): https://bitcointalk.org/index.php?topic=340157.0It is a second-generation currency, and should not be listed as just another *coin. Just like Ripple, eMunie, Protoshares don't have coin in their name. |
|
|
|
All I get when I run it from the windows cmd line is this:
D:\nxt\nxt>java -Xms512m -Xmx1024m -jar start.jar
2013-12-03 21:33:53.401:INFO:oejs.Server:main: jetty-9.1.0.v20131115
2013-12-03 21:33:53.433:INFO:oejdp.ScanningAppProvider:main: Deployment monitor
[file:/D:/nxt/nxt/webapps/] at interval 1
[2013-12-03 21:33:54.395] Nxt 0.3.7 started.
[2013-12-03 21:33:54.405] Loading transactions...
[2013-12-03 21:33:54.559] ...Done
[2013-12-03 21:33:54.559] Loading peers...
[2013-12-03 21:33:54.568] ...Done
[2013-12-03 21:33:54.568] Loading blocks...
[2013-12-03 21:33:55.330] 10: java.io.EOFException
2013-12-03 21:33:55.331:INFO:oejsh.ContextHandler:main: Started o.e.j.w.WebAppCo
ntext@1615ef2{/,file:/D:/nxt/nxt/webapps/root/,AVAILABLE}{D:\nxt\nxt\webapps\roo
t}
2013-12-03 21:35:22.345:INFO:oejs.ServerConnector:main: Started ServerConnector@
1384f2{HTTP/1.1}{0.0.0.0:7874}
Nothing actually happens? Been that way for 10 minutes. And was like it when I tried earlier.
Any ideas please
Looks like your blocks.nxt file has been corrupted. You will have to delete blocks.nxt (should be inside the nxt folder) and try again. If it still doesn't work, try getting the nxtfiles.zip file CfB posted earlier, which contains an archive of all *.nxt files up to some point, it contains blocks.nxt to help bootstrap the blockchain. |
|
|
|
|
I've registered an account on Nextcoin.org using Jean-Luc.
|
|
|
|
No need to be private at all, just to be sure that all stakeholders will be heard in big decisions ..
EDIT: And some better management of bounties, giveaways, promotion...
- I think nextcoin.org is good place for that. Already regestered there. Fine with me, except nextcoin.org still cannot be accessed over https. They should fix that. Yes, I am also a >1M stakeholder. |
|
|
|
The last version (0.3.3) deadlocks very often while trying to get the blockchain. After it hangs, Ctrl-C cannot kill it (because the thread is waiting to obtain a lock and ignores interrupts), so the only way to kill the java process is kill -9. Which leaves blocks.nxt in a corrupted state, so next time I try to start it, it fails with EOFException... not much fun. Looks like multiple threads are trying to update a HashMap (or several HashMaps) and deadlock one another. Would using a ConcurrentHashMap help? "qtp2032502559-269" prio=10 tid=0x00007fb9286dd800 nid=0x4b83 waiting for monitor entry [0x00007fb9041
a2000]
java.lang.Thread.State: BLOCKED (on object monitor)
at Nxt$Block.pushBlock(Unknown Source)
- waiting to lock <0x00000000f5e3ff28> (a java.util.HashMap)
at Nxt.doPost(Unknown Source)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:707)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:790)
I see quite a few of the above when sending a kill -3 to the java process. |
|
|
|
|
Show Posts
