So perhaps thatās a bug worth noting, if it hasnāt been noted before. Once a website folder fails to upload on one account, you need to create a whole new account to successfully upload a website folder.
I think I need to reset my proxy file link to get it working on my browser, was working fine last night but now firefox shows āProblem loading pageā
NOTE: fixed, by connecting through my own hotspot. I guess starbucks Wi-Fi doesnāt allow connecting to proxies like this
EDIT: nevermind, not working to view. Is it working for anyone else?
EDIT2: fixed again. I had to click this button in firefox to get the proxy working to view sites again. Hope it helps someone having the same problem:
arghh I hit the 100 puts after having deleted a file to update it, so for now, http://safe.share.safenet is screwed !
Also when I try to create a fresh account and a new name , I also hit lowbalance and uploads fail. I would have thought this is per account ? Or is it per IP ?
EDIT : note to self : DONāT upload from git repo, as it uploads .git too, thus hitting the 100 puts cap !
Problem :
I have accidentally my test account ! I didnāt think that uploading using the demo app would suck my .git folder, so my put limit went boom.
Once the account is locked , as it takes a ( refused ) POST to authorize with the launcher, I cannot auth and delete the long name, or delete the files so the public website doesnāt show in a broken state.
It would be good to be able to auth despite the cap limit, so one can make this kind of maintenance on a screwed account.
It would be good if the launcher sends you a warning before you hit the cap, too. Something like "If you do that, you would lock this account " , and the originating request would be ignored, so the account is still accessible for auth , get and delete stuff before it is really screwed.
Iām sorry, but how can that be accurate when itās so inconsistent? It gives me a different answer every time, so I canāt trust my trading to what it tells me.
Kidding of course! Nice page
Glad the stable droplet test network is indeed proving pretty stable so far!
I have connected 3 vaults (yes, I know I shouldnāt have) and added http://mystats.safenet/ site which display some statistics about the network.
The total number of vaults is estimated with empirical formula defined here, but 3 measuring vaults are not enough and only give a rough estimation.
Some averages on data in my 3 vaults are also displayed. Interesting extrapolations could be computed but uncertainty is very high. For example, the formula for the total number registered users would be: Estimated total vaults * Login packets / 8 (~ 365 users at the time of writing).
The site has no javascript but has an automatic refresh every 15s.
I wonder if thereās a way of getting at the network capacity through the amount of data a stable node has downloaded, multiplied by the estimated nodes. If the network was maxed out, how would that become apparent? Perhaps in future the capacity will evidence via the reward in safecoins??
Also, is it possible to have nodes attached to the stable droplet network, I thought only the launcher and demoapp had been made available and the community net is obviously distinct from that.
Before the great simplification we could guess if the network was about to run out of free space by looking at the sacrificial/normal chunks ratio. I donāt know if we can get the information at the vault level today. I will check that.
I cheated: compiled safe_vault from source and copied crust.config file of stable droplet test network.
Would that be fatal to an upload; or was I being impatient?.. the upload seemed not to be progressing beyond that error though I didnāt wait for more than a couple of minutes to see. I did wonder that the test network was being updated but then created a new account and uploaded without incident.
Going by what happens with vaults connecting to a network, it should wait a while and try again by another hard-coded contact, if thatās what is happening. But since this is a test network, you will probably have to take some action, such as trying again later.
I guess you mean ābootstrapā node. ātunnelā node has a different meaning in the network(its a node thats relaying messages to a given other node for the source since itās not able to connect to the destination directly itself).
In this case, hereās the source. Looks like the client has lost its connection to its bootstrap/proxy node(could be cos of multiple reasons, bootstrap node went offline(not likely). client connection terminated cos of local network fluctuating, ā¦).
This above message indicates a message such as this should have also got logged before in the log.
I donāt think its you being impatient and yes this would be fatal to the upload and would lead to the same issue you raised in GitHub I think. At this point, if the Put requests went through past the proxy node before its connection to your client terminated, then the network could have very well stored the chunks but not been able to indicate success to the client and thereby the metadata/directory listings for the file might not be updated accordingly(or even only part chunks could have been uploaded) and the rollback / recovery process isnt in place to just resume on retry right now.
Thats not something that happens for Clients. For clients the expected behaviour with such failure is for Launcher to have received a āDisconnectedā event and indicate to the user that they would need to reconnect to the network and Login again to resume their operations. For vaults its simpler as they just restart and create a new identity currently.
@davidpbrown If you still have the log file from that session and you donāt see a message such as Lost connection to last proxy node in your client log file, that would point to an issue where your node hasnāt realised it and informed launcher of the same.
Actually I meant tunnel because thatās what a proxy does, right? Iāve seen the proxy error many times, and itās usually because the hard-coded contact is offline.
It should not do that, it by default has to be directly accessible, to bootstrap from Tunnel is more like a network level proxy between network nodes, not client ā bootstrap nodes. In this case proxy is āacting on behalf ofā a tunnel node is ārelaying forā if that makes sense.
OK, thanks for the clarification, that thereās two meanings of proxy. It was the relay definition I already had, but the proxy error I was seeing (with client vaults joining by trying to connect to an offline vault in the config).