'chunks' of blocks not pasted/removed during an edit

[leasoncre]

v3.2.2
[same issue as previously mentioned somewhere..]
Paste action: a section of a schematic won't get pasted [appears like a missing chunk, may or may not be a full actual chunk] on the first paste. a second paste will paste the missing part(s) https://gyazo.com/4225792a3e008f75841d719659958e29
Remove action (usually a //set 0 for my uses): a section of the edit will get missed. ~about 1/8'th of a chunk, top of selection to bottom of selection. https://gyazo.com/48166c21fed8ae682f6a8c0839c265b0

i think this is somehow related to the world's chunk lines, because the 'section' not placed/removed correctly usually lines up with 3 of the 4 sides of the chunk. I know it's not a render glitch because i have tried re-rendering the chunks by logging and/or leaveing and comming back to the area. A server restart doesn't fix it either.

Related:? (some wierdness probably on my end....)
-when i paste/set, i leave the area (past the 10 chunk client render distance). this causes the operation to speed up alot since the server doesn't have to send updates to the client. and sometimes the operation will lag me alot untill i leave as well. server memory seems unaffected by w/e is lagging my client. and this particular problem is only experienced by me.

[SBPrime]

Any errors on the console? AWE does not touch the actual block placing its done by WE.

[leasoncre]

not that i noticed or found

[leasoncre]

still an issue with 3.2.3.
could this be somehow related to a chunk not being loaded by the server until the edit is part way through the chunk? [no players in vicinity to load the chunk, chunk already generated]

granted it's only a problem in (guesstimate) 10-15% or the chunks outside of the server's render distance.

[SBPrime]

This could be the case if you are using direct chunk commands not the regular commands. Are you using any block loggers or something that could cause this? Are you able to provide a step by step reproduction for the bug? Are you able to reproduce the bug on a clean server, reproduce it on production 100% of tries.

[leasoncre]

i don't know squat about what the chunk api thing does or really is. but it sounds like a cool thing.
i've never used any of the chunk commands aside from the /chunk relight one that worked with a regular selection box in 3.1.1.

test procedure:

1. generated new world on a fresh server (used the new one with only WE and AWE, that i used to test the relight command yesterday)
2. generate ~1000x1000 blocks worth of chunks (for the edit test)
3. //pos1 at ~80 blocks up (//sel extend mode)
4. expand to bedrock
5. expand 1000 s,e
6. //set 0
7. wait.
8. get DC'd
9. related: this problem may be related-to/caused-by when the server writes to disc during an async/WE operation. processor usage is at about 200% with async running a task. it spikes to 800% when the server writes to disc. (only way to see this was to run a server on a linux system and actively watch the utilization statistics.) -this was tested remotely on a friends server that we plan to move to in the future.
10. no chunk errors found. async works as expected
11. stop the server, and restart.
12. re-initiate the same edit, without loading the chunks previously generated on last startup (didn't move from where i spawned in) (they should be unloaded until a player enters them) (repeat 3-7)
13. get DC'd
14. no chunk errors found.
15. stop server, add coreprotect plugin and blocks hub, start server.
16. repeat 3-7, except with set to stone
17. had to do-over as i started selection in wrong spot, restarted
18. redo step 15 from right position.
19. get DC'd (server log says it crashed this time, but still running, console unresponsive) log: https://www.dropbox.com/s/3ts3xmtgwx1lu5f/Awe%20crash%20event.log?dl=0
20. no chunk errors found
21. stop server, restart (interesting: it looks as if one of the region files was not saved to disc (the one where i spawn into the world @ -382 83 130)
22. repeat 3-7: changes: //set 1, size 500x500 (still ~80 high)
23. get DC'd
24. no chunk errors found
25. stop, and restart server. added worldguard plugin (no reason why this should affect antything, but we'll see...) [[well.. this is getting difficult...]]
26. set 0 on 500x82x500 area from point of spawning in to the world.
27. get DC'd
28. no chunk errors found
29. stop/start server, added blockshub-WG plugin, forgot to add
30. set 1 on 500x500
31. get DC'd
32. ERROR FOUND: https://gyazo.com/6b96db86a0e2b21fa5e61e0c2826fdcc just the one error though. i usually see a couple. but not always. error may not be related to the blockshub-WG plugin (personal opinion) LOG: https://www.dropbox.com/s/cdt2aa5nm6magob/Chunk%20Error%20log.log?dl=0 error is approx 25-30 chunks away from player.
33. stop/start server
34. repeat //set 1 500x500, region shifted south to where last edit stopped (~122blocks)
35. get DC'd
36. no chunk errors found
37. stop/start server
38. repeat //set 1 500x500, region shifted south to where last edit stopped (~265blocks)
39. get DC'd
40. no chunk errors found

not testing this anymore today. it seems to be a very hard thing to replicate. but i can replicate it.

test world/server thus far: [[4gb file unzipped, 512mb-ish zipped]] https://www.dropbox.com/s/zsutswzd53r73hf/%C2%A4MC1.8.8fresh.rar?dl=0 [15 min left on dropbox file sync at time of post]

[leasoncre]

i can hazard a guess that these errors occur (both the DC's and the chunk issues) when tasks grab processor time and fight for the memory/cpu cycles. because we (currently) run our live server on a shared hosting machine, i can see this becoming rapidly evident. as i've spoke with the hosting staff and they say that there's 50 servers running on one machine.

It's possible that i incurred the one error on my windows-based server, buy doing other tasks on the machine while the edit was in progress. spike the machine's utilization..