V12 Release

I don't think the masternode stopping issue is RAM. I am running 256MB virtual machines running Ubuntu 64bit on my own server without problems. Linux uses almost all the system ram even if it doesn't need it. My own server nodes have not stopped with the new release or last release. The other nodes running on a remote server tend to crash every 5-15 days or so with last release. I believe it is because the CPU usage is over xx% for a x minutes and the server operator kills the instance. I tried using crontab, but somehow the remove server didn't allow it to run. I ended up making a script to try to start nodes every 10 minutes. Since dashd won't start twice, you can just keep trying to start it and it will only start an instance if it isn't running. See my script below. Copy and past into a start.sh file and chmod +x start.sh. You can run with ./start.sh but you would need to keep a ssh window open. I installed screen(sudo apt-get install screen) and run with screen -dmS /start.sh.

Keep in mind this doesn't ever stop. You need to be quick updating your nodes(I was able to update without shutting down), or kill the process before you update. To stop screen -X -S 123123 quit. 123123 is the process id you can get from typing screen -ls.

/home/ is the full directory for your dashd

#To Start: screen -dmS /start.sh
#To Stop: screen -X -S 123123 quit (find 123123 from screen -ls)
while true; do
/home/dashd
sleep 600
done

Or even better, might be to add -shrinkdebugfile. If your daemon stopped because of a large debug file, this would fix it. Might be more difficult to diagnose a different problem.
 
Last edited by a moderator:
oaxaca said:
This is better:

@reboot /root/dashd -shrinkdebugfile
*/20 * * * * /root/dashd -shrinkdebugfile
Click to expand...

no good on a running dashd
$ dashd -shrinkdebugfile
Error: Cannot obtain a lock on data directory /home/ubuntu/.dash. Dash Core is probably already running.
 
moocowmoo said:
no good on a running dashd
$ dashd -shrinkdebugfile
Error: Cannot obtain a lock on data directory /home/ubuntu/.dash. Dash Core is probably already running.
Click to expand...

Already running is what you want. No harm, no foul?
 
Notes for those with problems keeping nodes running. This helped me.
  • Delete the debug.log (mine were over 100MB, and it seemed to crash nodes in less than an hour until I deleted them)
  • Restart remote nodes once after they completed reindexing. ./dashd stop - wait and confirm process is stopped (30 seconds or so)
  • Start nodes from local wallet one at a time. Seems that when I did a start-many, only a few remote nodes took the command. I used start-alias instead.
  • Delete the remote node wallet if it won't stop or above doesn't work. (This worked for one of my nodes)
 
Solarminer said:
I believe it is because the CPU usage is over xx% for a x minutes and the server operator kills the instance.
Click to expand...

I don't think so because VULTR don't do this, I'm even mining of the VPS on VULTR and I'm using 100% of the cpu without problem.
And even on VULTR my daemon crash.

Edit :
o_O, just saw a .46 on dashninja
 
Lebubar said:
I don't think so because VULTR don't do this, I'm even mining of the VPS on VULTR and I'm using 100% of the cpu without problem.
And even on VULTR my daemon crash.

Edit :
o_O, just saw a .46 on dashninja
Click to expand...

Hmmm. Maybe it is just some instability with the remote server or dash .23/.45 client. Switched to a lower cost host about the same time as the previous release and had the problems. The start script worked for me so I haven't spent any more time figuring it out. I am pretty sure it isn't a ram problem since my home servers use less ram and don't stop.

On the old clients (pre .23) I was running a cpu miner too on Digital Ocean along with a node and didn't have problems. So DO wasn't stopping processes back then. Could be now though.

Maybe we have a fix with the .46 coming. 0.3% is what? 5 nodes. Must be some testing going on.
 
After months working away fine my 2 MN's on Vultr (upgraded to V12) died last night. Also having difficulties running the cli commands. Looks like it's not installed correctly "file not found" despite file existing in same directory as dashd. Any help appreciated.
 
Hello

As many others I have to say that something is wrong.
I have some "pairs" of MNs, running them on Vultr servers. One VPS is able to run 2 of them - CPU always been in 10% range and even the occasional spike leads to 50%.
In many months I had to occasionally restart one node or another. Usually, a node is able to stay up 5-6 weeks, but one was still running (>10 weeks). I think that maybe a VPS may hold even 3 of them.

After updating everything, my nodes keep to go down after 5-6h, 24h is top record till now.
Vultr control panel doesn't show any spike of activity, they just go down. Usually in a pair one falls and the other stays up, but after a while I can also reach 0/2 so I really don't think it's a matter of resources.
BTW even in this case it would be a mess cause it would double the fixed cost - right now a node yelds 15Dash/m (i'd say 14 or even less if they go down...), If i have to pay 3 Dashes just for the VPS it's a serious cost I really don't like cause it's cutting my eventual profit to a range where I don't make money for the risk I am taking here.

I really hope we get updates, I restarted every node at least 5 times in 2 days (and I didn't get any dash yet since upgrade...) or I'll have to evaluate to quit

Cheers
 
Gi01 said:
Hello

I really hope we get updates, I restarted every node at least 5 times in 2 days (and I didn't get any dash yet since upgrade...) or I'll have to evaluate to quit

Cheers
Click to expand...

Don't worry, Dash dev team (udjin, evan, flare, etc...) is the most reliable and fastest in all alt ecosystem. We had a lot of issues in the past (this is edge tech, it is normal) and dev team always responds in hours.
 
I would recommend the people who are having problems to compile the .46-version from the source (it's been committed to master source)
what I have observed it working on my masternodes (half-a-day), lower CPU/MEM/Net-usage, without the sawing of these we saw earlier..

edit: seems like 11.5% (and climbing) is already in that version.

You can clearly see the difference at the end... (around after 7AM) with .46-version.
3444dah.png

2a0ic84.png
 
Last edited by a moderator:
256MB with no swap available seems to be too small for me to perform a reindex.
The best I can do is 30 to 50 percent progress and dashd crashes every time the daemon hits 95 percent ram usage. I've been playing around with ulimit, trying to find a variable that might actually limit dashd's memory usage temporarily so it can perform a successful reindex. No luck so far.
 
Bridgewater said:
256MB with no swap available seems to be too small for me to perform a reindex.
The best I can do is 30 to 50 percent progress and dashd crashes every time the daemon hits 95 percent ram usage. I've been playing around with ulimit, trying to find a variable that might actually limit dashd's memory usage temporarily so it can perform a successful reindex. No luck so far.
Click to expand...

have you tried getting the bootstrap?
https://github.com/UdjinM6/dash-bootstrap
 
Good morning.

2 more crash this morning and always with this last message in debug.log :

Code: 
Verifying mncache.dat format...
 
Is .46 up and running or still testing? I might wait some more days if we'll have .47+
On the other side, it's 50% less CPU, but i never crashed due to CPU spikes. I assume it's something else; whoever upgraded to .46 noticed a different behaviour?
 
Does it matter that the miners aren't updating? Or do the masternodes have the ability to reject mining blocks, so that they eventually have to comply?
 
You must log in or register to reply here.
Back
Top