Fix bug #80024: prevent duplication of inherited sockets on reload#6550
Fix bug #80024: prevent duplication of inherited sockets on reload#6550syazov wants to merge 1 commit intophp:masterfrom
Conversation
|
The change looks good. I have been just thinking that it would be worth to create a test for that as you provided nice way to recreate shell and think we could do the same in our tests. It already supports multiple pools but we would probably need to add in place config modification which shouldn't be that hard either. Then it would be just sending the signal and expecting start. It might need some further tweaking with output checking and not sure if it's gonna run in CI but would be nice to have some way to test it. I might give it a try and if it doesn't work, I will merge it as it is. Or if you want to try, that would be of course great. |
|
Just an update that I'm doing some work on extending multi pool support in the tester. Currently I'm working on the FPMi next branch which I often use for new things before I port them to FPM. The progress can be seen here https://github.com/bukka/fpmi/compare/next . |
|
This is covered by #6740 which also contains a test and a bit of clean up. |
"FPM_SOCKETS_X" env variable isn't cleaned when it's become unused.
The issue appears if count of sockets is changed from "FPM_ENV_SOCKET_SET_SIZE * n + 1" to "FPM_ENV_SOCKET_SET_SIZE * n"
In such cases the same socket presents in "FPM_SOCKETS_[socket_set_count - 1]" and "FPM_SOCKETS_[socket_set_count]"