r8845 bus-server.pl spawned THOUSANDS of times.
Posted: 16 Mar 2014, 21:37
There is something in the change to either the bus-server.pl in revision 8845, or the code that calls it that is clearly broken for me:
Backgrounder:
killall perl
Killing each window running openkore in screen, because it no longer displays anything shown on screen by kore, nor in the shell, after running killall perl.
Normally, I'm able to run these instances for 6 or 7 days up.
After the 3rd time noting this issue, I paused and checked the process list, using "ps -A".
This is a SMALL exerpt of a "ps -A" listing on the ubuntu machine in question.
Read this carefully: It is *5172 lines long*, of which < 100 other processes are NOT perl.
Something about the ability to allow multiple Bus servers that revision 8845 enables is doing a tad more than that. It's spawning approximately 1,200 *extra* instances of perl running the bus-server.pl script per instance of openkore run.
I have reverted 8845 out of my local code base and will test further to see if the problem still occurs, or not. Can the coder responsible for this commit check over their contribution, and see if they can spot the change that causes a plethora of bus-servers to spawn?
Backgrounder:
- Running openkore, revision 8848, updated via SVN
- Running Ubuntu
- Running 4 instances of openkore in separate screen windows, using the Console interface.
- All 4 loading iMikeLance's busCommands plugin (but not really using it).
killall perl
Killing each window running openkore in screen, because it no longer displays anything shown on screen by kore, nor in the shell, after running killall perl.
Normally, I'm able to run these instances for 6 or 7 days up.
After the 3rd time noting this issue, I paused and checked the process list, using "ps -A".
This is a SMALL exerpt of a "ps -A" listing on the ubuntu machine in question.
Read this carefully: It is *5172 lines long*, of which < 100 other processes are NOT perl.
Code: Select all
2492 ? 00:01:47 perl
2496 ? 00:00:00 perl <defunct>
2497 ? 00:00:00 perl
2510 ? 00:00:00 perl <defunct>
2511 ? 00:00:00 perl
2513 ? 00:01:32 perl
2517 ? 00:00:00 perl <defunct>
2518 ? 00:00:00 perl
2519 ? 00:00:00 perl <defunct>
2520 ? 00:00:00 perl
2522 ? 00:01:51 perl
2526 ? 00:00:00 perl <defunct>
2527 ? 00:00:00 perl
2528 ? 00:00:00 perl <defunct>
2529 ? 00:00:00 perl
2530 ? 00:00:00 perl <defunct>
2531 ? 00:00:00 perl
2532 ? 00:00:00 perl <defunct>
2533 ? 00:00:00 perl <defunct>
2534 ? 00:00:00 perl
2535 ? 00:00:00 perl
2536 ? 00:00:00 perl <defunct>
2537 ? 00:00:00 perl
2538 ? 00:00:00 perl <defunct>
2539 ? 00:00:00 perl
2540 ? 00:00:00 perl <defunct>
2541 ? 00:00:00 perl
2542 ? 00:00:00 perl <defunct>
2543 ? 00:00:00 perl
2544 ? 00:00:00 perl <defunct>
2545 ? 00:00:00 perl
2546 ? 00:00:00 perl <defunct>
2547 ? 00:00:00 perl
2548 ? 00:00:00 perl <defunct>
2549 ? 00:00:00 perl
2550 ? 00:00:00 perl <defunct>
2551 ? 00:00:00 perl
2552 ? 00:00:00 perl <defunct>
2553 ? 00:00:00 perl
2554 ? 00:00:00 perl <defunct>
2555 ? 00:00:00 perl
2556 ? 00:00:00 perl <defunct>
2557 ? 00:00:00 perl
2558 ? 00:00:00 perl <defunct>
2559 ? 00:00:00 perl
2560 ? 00:00:00 perl <defunct>
2561 ? 00:00:00 perl
2563 ? 00:05:24 perl
2566 ? 00:00:00 perl <defunct>
2567 ? 00:00:00 perl
I have reverted 8845 out of my local code base and will test further to see if the problem still occurs, or not. Can the coder responsible for this commit check over their contribution, and see if they can spot the change that causes a plethora of bus-servers to spawn?