that means the latest record to be quoted is an existing one.
maybe the vdash shall skip zero quote, as it only means the record is already presented.
a restarted node has to be in sync with his quoting history, i.e. use the similar quoting metrics (records stored, payment received, etc.), otherwise it will be considered as bad by others.
i.e. a restart won’t help you to reduce your quote
Would it be possible to give the equations/algo used for store cost in the OP of any test/beta network please.
Or at least tell me where in the code (on github) to grab it.
I am sure it’ll answer so many questions people eventually ask in every test net topic. Including this one
it’s the function calculate_cost_for_records
Which source file is it in or is that the name of the file too
safe_network/sn_networking/src/record_store.rs at main · maidsafe/safe_network · GitHub <<< found it here
I tried that start up technique in the past 
Still trying to figure out a number and speed that doesn’t end up in a mess
That is very true and I believe that is what VDash is doing - showing the last value - even if it is from ages ago. But just going by that is better than nothing if you want to stop the nodes that are likely more expendable to the network and likely to earn less.
I have been taking a different approach to telling it how many nodes it needs to run.
Write a script that starts nodes only until resource limits are met. That way you don’t exceed it with some magic number.
Then if some machines are shorter on muscle than others they start less than those with a bit more oomph.
hmmmm - i think the only unknown that remains is the received_payment_count … which should equal the number of cash notes in /var/safenode-manager/services/safenodeX/wallet/cash_notes/ …
sounds possible to do … the most annoying thing would be that it will be hard to stay in sync with maidasafes code base …
Nice tip, I shall do that.
On restart, I of course meant start a new node so the issue remains.
You are probably not going to like what that does for your numbers though, but better healthy than shunned, no?
I am having a hard time wrapping my head around how Maidsafe have 40 nodes on 2vcpu instances.
I did a little experiment and started 50 on a 4vcpu instance the load was peaking at twice what it could handle.
So now I am testing a digitalocean droplet with the same specs as Maidsafe’s with the same amount of nodes (40) I will be very surprised if they are not overloaded too.
So question @qi_ma does a queue at the cpu affect bad node detection? I imagine it causes delayed response?
you get that with
safenode-manager balance
What is unknown is
The node keeps inactive records in order to save on retrieving records from other nodes if they need them due to another node leaving. This reduces overall churn traffic, and I also suspect in future will provide some records for caching.
The node will delete a inactive node when it needs to store a chunk that it is a close node for. This way the record store remains less than or equal to 2048 records.
So we need active records in order to give knowledge on actual records costing will be worked out on.
was thinking about that but I think you were right before run less than you think is possible as the nodes.
id quite like to find a set and forget number that provides good nodes to the network not a horror show.
Anyone know if there is a way to select a particular hard drive when spinning up nodes through vdash? Thanks
think --data-dir-path might be just the ticket
Usage: safenode-manager add [OPTIONS]
Options:
--count <COUNT> The number of service instances
--data-dir-path <DATA_DIR_PATH> Provide the path for the data directory for the installed node.
--env <env> Provide environment variables for the safenode service
--home-network Set this flag to use the safenode '--home-network' feature
--local Set this flag to launch safenode with the --local flag
--log-dir-path <LOG_DIR_PATH> Provide the path for the log directory for the installed node.
--metrics-port <METRICS_PORT> Specify a port for the open metrics server
--node-port <NODE_PORT> Specify a port for the safenode service(s)
--path <PATH> Provide a path for the safenode binary to be used by the service
--first Set to indicate this is the first node in a new network
--peer <multiaddr> Peer(s) to use for bootstrap, in a 'multiaddr' format containing the peer
ID [env: SAFE_PEERS=]
--rpc-address <RPC_ADDRESS> Specify an Ipv4Addr for the node's RPC server to run on
--rpc-port <RPC_PORT> Specify a port for the RPC service(s)
--url <URL> Provide a safenode binary using a URL
--user <USER> The user the service should run as
--version <VERSION> Provide a specific version of safenode to be installed
-h, --help Print help (see more with '--help')
Can we not set how many logs to keep with it?
If we are gping to be running long term it will be needed, @chriso 
yes, could be.
if the delayed response is so slow that considered as timed out by the peer.
that will then be in a different close_group. the chance of gain more, or even less is still even, and may not worth to carry out a restart.
Bäm! 23 full token earned by farming!
Boom 42 SNT this node is earning it’s keep if I could just have 5 like it I’ll shut the rest down
I’m going for a staged shut down then going to go to the drawing board to try and come up with a resource based start up script