so I was able to sign up again using my initial account details, and also re-reg “wes” handle with the network. So all of my data from that account must have been lost in churn (I’ve always had at least 5 vaults running, even with my mass-crash) Either the network isn’t fast enough to handle our little network worth of churn (smaller network makes it harder) or the artificial caps imposed for testing purposes kept churn from happening.
New thread? (20 char)
On my first attempt at logging in I tried to create a new account with the same credentials and the console reported that the data already existed. So I then logged in with the same credentials. Then I had to recreate my username and website. So it appears that my login credentials survived the crash but nothing else did.
I think all credentials survived the mega crashed. It is the chucks that has been completely ruined.
Maybe an executive summary here, and if anyone needs more detail then they can ask?
I’d do both: a summary here and a new thread with a general title along the lines of “ways to crash SAFE nets” that will, in the months to come, accumulate many examples for the core devs to pore over. 
1 Like
I think you may be surprised just how many tests we do and what we find out 
for sure trying to crash the various test will be great, but we try and publish limitations, to breaking the limitations should crash them in many cases.
Great work though, we should find out more info I think from this kind of testing. A good idea is to try and grab the log files if you see issues and collate them. These would help us if we have a description of any issue you see.
In this case the issues are known and hopefully resolved for now, but we still wont resolve data loss in huge churn, just yet. A couple more tweaks and tests and we will though.
10 Likes
Which begs the question how to reboot from scratch… I guess in the current version, if anyone has a vault running it will play host to legacy data and confusions. Perhaps need the next version with network name in order to have option to declare a new network and let others know it, as there’s no way to be sure old vaults are not still running.
This is exactly what is being prepared by Maidsafe. So I have corrected my older post:
1 Like
I see no drawback. There might be cases where you don’t want to reject old nodes. If the the version number was used that would complicate adding newer type nodes to an existing network. New networks will necessarily now declare themselves with a unique name and obviously a versioning in the name will be sensible… unless we go all Ubuntu like and spawn names from letters of an alphabet. Aardvark 1.0 can see then node versions 0.01 and 0.02 etc.
1 Like
I haven’t tried to use the launcher and demo app since last night. I used them just now and got in quickly once I used new credentials (I didn’t try to re-register with the old credentials, once i found that they did not work for logging in). I registered my username again, and re-created a website. So, it seems that in the last 18 hours some collapse occurred that lost everything.
Here is my current offering:
That plot is totally fictitious data. The nice thing about it is that it is an SVG file (i.e., a vector graphic), compact and ideal for web graphics, and generated programatically from a text file of data (that I made up). The software used to create it is Gnuplot.
I am amazed that this testnet works right out of the box on windows 10 /64. It is very promising.
Just followed the procedures and I could create an account and create a website test.hans.safenet and upload a file. I also could see some safe-sites int this forum thread.
Next day, I couldn’t log in anymore and I needed to make a new account. test.hans.safenet didn’t come up anymore.
Next day same story with another safenet-site and account.
Now I’m using a third account, with new site http://test.hans2.safenet . It works at this moment.
But I cannot connect to other sites anymore: always get this error:
{“errorCode”:-18,“description”:“CoreError::GetFailure”}
Question: When I start up my safe-vault I always get this Warning below
Do others get this too? Or is it something in my config or my proxy ? (safe_proxy.pac)
INFO 00:13:50.567330000 [safe_vault::personas::data_manager data_manager.rs:314] Stats : Client Get requests received 0 ; Data stored - ID 7 - SD 13 - total 2928254 bytes
WARN 00:13:50.786298700 [routing::core core.rs:598] Node(e49f…) Failed to connect to peer PeerId(e269…): Error { repr: Custom(Custom { kind: TimedOut, error: StringError(“Connect failed. errors: (1 of 5) Tcp direct connect failed: #The requested address is not valid in the context of the address# (os error 10049) (2 of 5) Tcp direct connect failed: #The requested address is not valid in the context of the address# (os error 10049) (3 of 5) Tcp direct connect failed: Zero byte read - EOF reached, graceful exit. (4 of 5) Tcp hole punching failed: Tcp hole punching timed out without making a successful connection. The following warnings were raised during hole punching: 25 warning(s): (1 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (2 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (3 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (4 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (5 of 25) Connecting to endpoint 127.0.0.1:58848 failed: #Cannot make connection because the targetcomputer has refused the connection actively # (os error 10061) (6 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (7 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (8 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (9 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (10 of 25) Connecting to endpoint 127.0.0.1:58848 failed: #Cannot make connection because the targetcomputer has refused the connection actively # (os error 10061) (11 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (12 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (13 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (14 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (15 of 25) Connecting to endpoint 127.0.0.1:58848 failed: #Cannot make connection because the targetcomputer has refused the connection actively # (os error 10061) (16 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (17 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (18 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (19 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (20 of 25) Connecting to endpoint 127.0.0.1:58848 failed: #Cannot make connection because the targetcomputer has refused the connection actively # (os error 10061) (21 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (22 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (23 of 25) Connecting to endpoint 0.0.0.0:34387 failed: #The requested address is not valid in the context of the address# (os error 10049) (24 of 25) Connecting to endpoint 96.237.141.18:58848 failed: #Each sockerAddress (protocol/networkAddress/pot) can normally be used only once#(os error 10048) (25 of 25) Connecting to endpoint 127.0.0.1:58848 failed: #Cannot make connection because the targetcomputer has refused the connection actively # (os error 10061) (5 of 5) Tcp direct connect failed: #An connectionattempt has failed because the connected party has not answerd correctly after a certain time, or the connection has failed because the connected host has not answered # (os error 10060)”) }) }.
1 Like
note: part of the above warning is translated from Dutch with possibly some errors.
sockerAddress = socketAddress.
1 Like
Just gave this a go. I started a vault and the safe launcher. I could create login credentials and then launched demo app. So far so good. Uploaded the example site and set the proxy. I could then load my own site in firefox, but I have not been able to load any other sites from this thread. Some will just show a “Connecting” tab where it waits for ages for a site to load, but nothing ever happens. Other sites will just give me a short line of code stating an error occurred opening the site.
When I tried to edit my site via the demo app it seemed the connection was lost. I then couldn’t even load my own site anymore. Restarting the launcher and logging in again made me see my own site again, but still no others.
As for the vault it’s been taking 12 to 16 percent CPU. I have an old laptop with a dual core i5, 8gb of RAM and I’m running Ubuntu (Mint).
Welkom Twiet 
This is a small testnet started by the community. We’ll see a testnet from Maidsafe in the coming days is my guess. That should be more stable after the test we did about 10 days ago. The current test is maybe 30 or 40 nodes. Last testnet from Maidsafe had 790 nodes (590 from community). Next test should be way bigger and way more stable.
2 Likes
I logged in just now with my previous credentials but the demo app could not find my username (with activity animation going for ten minutes). I created a new account and tried to register the old username but the error message is that it is already taken. So i have created a new username and website. Everything new happens very quickly.
I will run four vaults today to see if that mitigates this apparent data loss through churn or network collapse.
By the way, the reason the network is much smaller than during the official tests is that, in general and not just with this group, people have a slave mindset, and their default approach to life is to wait for authority figures to tell them what to do. This is why most people equate livelihood with “getting a job,” for example, and to get their model of the world from the mass media or the government.
2 Likes
The main issue is that new data is lost or corrupted after 12-18 hours. That is a repeating pattern. There is generally no one event that I could point to as the cause.
The priority of loss appears to be: websites and usernames become inaccessible, and lastly login credentials no longer work. But even if login credentials and usernames are no longer usable/accessible they will be reported in error messages as already taken. Only with a complete collapse, such as at the end of the first day, is the slate wiped clean and one can begin over again by reusing the same credentials.
The network is distinguished by two factors: It’s small size and the fact that there are only three “hard-wired” nodes in its default config.
However, I’ll collate data from my logs if you let me know what you might find useful.
3 Likes
It seems even less reliable, in using the launcher and demo app, with four vaults on my dedicated server.
I’ll stop the vaults and restart with only one vault.
EDIT: Once again, everything is fast after a restart of the vault at 91.121.173.204:5483
The other two hosts in the config file do not respond to a telnet request, so they are not listening on that port. So there is just the one machine listening, which evidently isn’t enough for much of an uptime. It seems to get “tired and confused” after a few hours.
Do you (or anyone else out ther) have an IP of a vault that I could add to the config file?
Yup I just noticed that too. I was about to make a comment about that. Now you have, I don’t have to add same information. side note, I can’t access to my new website after it was freshly created Saturday night.
So I guess it’s not the rsafe program that’s causing the issue. It just happened to be on point when corrupted vaults started to happened. It was mere coincidence. I was skeptical how the rsafe corrupted the entire safenet but after two days of research…I honestly don’t think so.
My routing table is going down to only 9 nodes. So I think our little testnet is crashing.
EDIT: Yup, network down. My Vault restarted itself.
OK, I’ve taken the vault down, since I feel that we have learned as much as we can with just one valid hard-wired node in the config. If anyone wants to volunteer an IP I’ll run it again.
What I would like to know is what is the minimum number of hard-wired IPs or size of network that will keep it running and fast permanently, or on a sliding scale from better to worse over longer times as those two parameters are increased?
Suitable tests would be to keep one parameter fixed and increase the other, and periodically record response times and failures to login, or retrieve data, etc. That’s going to take a lot of testing to arrive at an equation that succinctly expresses those relations.
I understand that Maidsafe do such testing. Is it published anywhere?
3 Likes