Dry Run 8: Need for Speed

[bytemaster]

I just now restarted my wallet with 10 delegates .. however .. during the downtime several block have been produced by the network .. once I restarted my wallet (same round I guess) my wallet produced a block although not syncronized to the network leaving the missed but produced blocks as MISSED. I guess I thus ended up on a fork .. may jusg be a minor issue though ..

Now i restart the wallet and let it sync before unlocking the delegates keys

This is a best practice and something we should automate. 


Sent from my iPhone using Tapatalk

[betax]

Just got down to 4 connections, quit and restart then 8 connections straight away. I guess is as simple to keep an eye for the time being.

[xeroc]

I just now restarted my wallet with 10 delegates .. however .. during the downtime several block have been produced by the network .. once I restarted my wallet (same round I guess) my wallet produced a block although not syncronized to the network leaving the missed but produced blocks as MISSED. I guess I thus ended up on a fork .. may jusg be a minor issue though ..

Now i restart the wallet and let it sync before unlocking the delegates keys

[sumantso]

Anybody?

Before you begin make sure your block production is off:

wallet_delegate_set_block_production sumantso true

You will need to then do

wallet_approve_delegate sumantso

do a transfer to broadcast the vote

wallet_transfer 10000 XTS sumantso sumantso

and then re-enable block production

wallet_delegate_set_block_production sumantso true

My one delegate is already producing blocks. I tried setting up a 2nd one, after turning off block production, but it keeps saying insufficient funds.

[sumantso]

Ok, got the moolah for a second delegate, but it doesn't allow to register saying insufficient funds. Any ideas?

I had disabled block production for that time, or do I need a separate client altogether?

Anybody?

Don't need a separate client, just the funds to register. Once is registerd vote for it and re-enable block production for both.

Not happening. Tried the trick in various ways.

[bytemaster]

Incremental update with various minor fixes to the UI.

Mac OS X: http://bitshares.org/BitSharesXT-0.1.1.dmg

[bytemaster]

We have identified the cause of the connection issues:  memory leak that is causing us to not free up file handles on the sockets.    Now that we know the problem it is now just a matter of finding out why it is occurring.

In the mean time that means delegates should probably reset their node a couple of times per day to maintain connectivity for everyone.

This will likely be fixed tonight or tomorrow.

FWIW - mine has been running fine on Windows now and only using 167MB of RAM. (Granted I have 32GB of RAM, but it's not even close to using it).
So I am not seeing evidence of a memory leak. However on the file handles - that is quite possible. Process explorer was showing peak of over 2000 handles (but I think that includes GDI handles)

I think it applies more to the public nodes not behind a firewall, but that is just a theory.

[bytemaster]

Is it possible to make the number of active delegates be adjusted by the community via voting?

I personally feel that as the network grows 100 delegates would be too small and in fact will look like "centralized power" to the rest of the world.

This has been discussed in other threads.  It is a very complex topic.

[bitmeat]

Anybody?

Before you begin make sure your block production is off:

wallet_delegate_set_block_production sumantso true

You will need to then do

wallet_approve_delegate sumantso

do a transfer to broadcast the vote

wallet_transfer 10000 XTS sumantso sumantso

and then re-enable block production

wallet_delegate_set_block_production sumantso true

[betax]

Ok, got the moolah for a second delegate, but it doesn't allow to register saying insufficient funds. Any ideas?

I had disabled block production for that time, or do I need a separate client altogether?

Anybody?

Don't need a separate client, just the funds to register. Once is registerd vote for it and re-enable block production for both.

[sfinder]

We have identified the cause of the connection issues:  memory leak that is causing us to not free up file handles on the sockets.    Now that we know the problem it is now just a matter of finding out why it is occurring.

In the mean time that means delegates should probably reset their node a couple of times per day to maintain connectivity for everyone.

This will likely be fixed tonight or tomorrow.

tiger (unlocked) >>> socket: Too many open files
socket: Too many open files

[betax]

We have identified the cause of the connection issues:  memory leak that is causing us to not free up file handles on the sockets.    Now that we know the problem it is now just a matter of finding out why it is occurring.

In the mean time that means delegates should probably reset their node a couple of times per day to maintain connectivity for everyone.

This will likely be fixed tonight or tomorrow.

That explains why mine drop connections after ~ 13 hours.

[sumantso]

Ok, got the moolah for a second delegate, but it doesn't allow to register saying insufficient funds. Any ideas?

I had disabled block production for that time, or do I need a separate client altogether?

Anybody?

[bitmeat]

We have identified the cause of the connection issues:  memory leak that is causing us to not free up file handles on the sockets.    Now that we know the problem it is now just a matter of finding out why it is occurring.

In the mean time that means delegates should probably reset their node a couple of times per day to maintain connectivity for everyone.

This will likely be fixed tonight or tomorrow.

FWIW - mine has been running fine on Windows now and only using 167MB of RAM. (Granted I have 32GB of RAM, but it's not even close to using it).
So I am not seeing evidence of a memory leak. However on the file handles - that is quite possible. Process explorer was showing peak of over 2000 handles (but I think that includes GDI handles)

[CalabiYau]

We have identified the cause of the connection issues:  memory leak that is causing us to not free up file handles on the sockets.    Now that we know the problem it is now just a matter of finding out why it is occurring.

In the mean time that means delegates should probably reset their node a couple of times per day to maintain connectivity for everyone.

This will likely be fixed tonight or tomorrow.

Code: [Select]

calyauwall (unlocked) >>> socket: Too many open files
socket: Too many open files
seems to be the right track - looking forward