Separate names with a comma.
Discussion in 'Proof-of-stake Mining' started by rohit pawar, Feb 19, 2016.
As i understand i don't need static ip for participating in "PoS mining". Am i right?
yes that would be correct. You need static IP for running a fullnode i think. A fullnode is a dcrd that uploads old blocks to new people syncing for first time.
When you update your binaries at AWS do you need to stop your POS?
I wonder if there is any way of doing without stopping my POS as normally it takes up to 30 minutes to sync a new version and don't want to miss voted tickets.
@Adriano, yes, you need to restart dcrd and dcrwallet. Try to stop dcrd daemon using this command:
It will take some time to save db, but after this you can start it in some minutes.
This command will not stop the wallet, if I remember well.
In Windows Ctrl + C will also result in a shutdown yes/no prompt. This is a controlled shutdown. Killing the window is not. I had not realized this until someone mentioned this somewhere else.
If I don't stop the wallet my ticket can still be voted?
Just copy blockchain into a new dir
Free AWS usually can stand 2 daemons and 2 wallets running.
How long to download the whole blockchain using a AWS windows?
On my computer it goes fast but at the AWS it is going very slow.
If it is a t2.micro (free tier), it may use up all your CPU credits. This will make it slow to a crawl.
BTW, 2 daemons on a t2.micro is too much. It will likely OOM.
So what would you recomend at aws windows to have to go faster with 1 daemon?
Should I use t2.small, medium or bigger to make it faster?
I prefer windows as I'm not so familiar with linux.
@Adriano A t2.small is obviously better as it has more adequate ram (2GB), and double the CPU credits, but it still relies on burstable cpu so no guarantees. Of course, they make you pay for it.
Regarding OS, you can put Windows Server on any instance.
But my recommendation would be to try Google Compute Engine. Their g1-small is great and well priced.
Thank you for your advice and patience. Had a look about it and understood about the credits. That was the reason it was taking so long.
Have upgraded to a t2.large and it is much faster already with more credits.
It's worth to pay due to the speed.
For POS the more RAM the better? If I would be looking for a better instance these is what I shall look for?
There are so many instances that it get confused.
Have a look at CloudWatch to see CPU credits over time. They plummet quickly.
RAM is needed for dcrd, not PoS per se. Once you are synced, you don't need much compute power. But if you are out of CPU credits, expect your vote to be late. What happens is:
dcrd processes the newly mined block
dcrwallet gets the notification from dcrd
dcrwallet processes the block and realizes it gets to vote on the block
dcrwallet creates and transmits through dcrd a SSGen, the send stake transaction, a.k.a. the vote
You don't need a super computer to do this, but it has to be responsive and reasonably fast.