[bug] I think I crashed the entire Community Test Network

Development
1

Continuing the discussion from Join the Community Test Network (Running Now):

I ran the rsafeclient example. It connected to the launcher no problems. I left it for couple more minutes before I ended the example test. After I ended. One minute or so, the launcher crashed. Here are the errors on the launcher terminal.

/build/Safenet/safe_launcher-v0.4.3-linux-x64/resources/app.asar/node_modules/ffi/lib/callback.js:18
throw new Error(err)
^

Error: ffi fatal: callback has been garbage collected!
at /build/Safenet/safe_launcher-v0.4.3-linux-x64/resources/app.asar/node_modules/ffi/lib/callback.js:18:15
at doNTCallback0 (node.js:443:9)
at process._tickCallback (node.js:372:13)
Griz@LocalHost:~/…/safe_launcher-v0.4.3-linux-x64$ ./safe_launcher
WARN 16:42:39.673459218 [w_result lib.rs:288] Error mapping external address and port through IGD gateway at address 192.168.0.1:49155. igd::Gateway::get_any_address returned an error: Request error. Gateway response error 401: Invalid Action
WARN 16:42:40.423329412 [routing::core core.rs:1327] Connection failed: Proxy node needs a larger routing table to accept clients.

WARN 16:54:06.691321743 [w_result lib.rs:288] Error mapping external address and port through IGD gateway at address 192.168.0.1:49155. igd::Gateway::get_any_address returned an error: Request error. Gateway response error 401: Invalid Action

| FfiError::DnsError → DnsError::CoreError → CoreError::GetFailure::{ reason: NoSuchData, request: Structured(7ee8…, 5)}

thread<‘unnamed’>panicked at ’

======================================================
| Result evaluated to Err: “Failed decrypting chunk 0” |
======================================================

Then Vaults started to drop all of the connections. After 5 minutes or so, vaults are able to connect again. Unfortunately, the vault errors has been completely overwritten by new data so I can’t copy and paste the vault incident during that time frame. But mostly, it said TCP can’t connect.

And then the community said they can’t access to the website or the files. Yikes.

In all of this madness, I believe the script has completely deleted or override the files of the entire network. Thus, nobody can’t access to the websites at all. The vaults seems to be working now but all of the files are lost, and needs to be rebuild again. The churn event couldn’t handle the amount of lost data.

3 Likes
2


7 Likes
3

uh oh !

I hope rsafe is not faulty in this… I wouldn’t imagine it doing weird stuff at the network level as it basically only talks to the launcher , but well, it would be interesting to sort that out and make sure it doesn’t happen again.

By the way, rsafe is outdated , I made some progress since then, but I didn’t push to github yet , as my current version is screwed, and I couldn’t work on it recently due to moving to a new home. I hope to be able to work on it again soon.

@Audity lmao

4

My vault log might me usefull,
Time is gmt +2.

3 Likes
5

Warning, really long vault log. I think around halfway to 3/4th is when the crashed happened.

That’s what I assumed. I think nfs over write might be at fault here. All of the chucks are no longer there or been overwritten. It cannot decrypted the chuck files.

1 Like