BakermanLP: I've updated the trigger feature to allow capturing logins, see version 1.50 and "/sampler trigger login". It'll capture every login, but the first number in the filename has the spike length in ms, so you should eventually see a problematic one. I'd like to have a look at the nps file if you manage to obtain a sample for a long pause.
I've tried the new sampler version. Here are my results:
- Server timing after client has started: 673 ms ( File: http://projectret.de/files/login-673-23191181704.nps )
- Server timing at next login: 23ms ( File: http://projectret.de/files/login-23-23191583546.nps )
Client (via log messages);
- Client timing after client has started: 08:33:10 to 08:33:56 = 46 sec
- Client timing at next login: 08:40:03 to 08:40:22 = 19 sec
My results so far:
- Server timing doesn't seem to be a problem
- Client timing seems to be a problem, but this isn't mesureable in any way, or?
Additional information:
- Modpack: https://github.com/BakermanLP/ProjectRETv8
- Forge 1492
Logfiles (Client):
- First login after clean restart of client: http://pastebin.com/wLra7LWE
- Client Log next login (no client restart): http://pastebin.com/H52WY6n6