Keery Nie

We're using ubuntu:20.04 in github actions test environment, which already has tzdata installed https://github.com/actions/virtual-environments/blob/main/images/linux/Ubuntu2004-Readme.md#installed-apt-packages

+1 for this, we need URI captures exposed in PDK as a convenient and clear interface that can be used in plugin code

This is a nice-to-have feature for rate-limiting plugins, but it may require more work to be done, the feature itself relies on the existence of the timezone database(in most Linux...

>a dependency of a dependency I believe it is the truth. So I'm checking whether we can remove all of these(including the globalpatch in EE), since we already solved the...

Oh, I forget that cosocket is also not available in init_phase, so resty.http cannot be directly used ``` 2024/02/29 21:22:56 [error] 26418#0: init_by_lua error: ./kong/globalpatches.lua:574: no request found stack traceback:...

Single service creation test Before: ``` 2024/01/09 22:16:00 [debug] 15117#0: *236 [lua] init.lua:228: process_event(): [cluster_events] new event (channel: 'invalidations') data: 'services:c9ff5ac9-6833-4ba2-be5d-460c8aad47b5:::::ab2e831b-d916-4389-a714-caf1e558c300' nbf: 'none' shm exptime: 10 2024/01/09 22:16:00 [debug] 15117#0:...

>Given this, things we could do for now (patch): >Leave the existing channel only for kong.cache -- this is what most (> 95%?) of plugins want when they do :invalidate...

@gszr @flrgh Please reivew again, thanks! I'm now following GS's advice on leaving old channels for db_cache and creating a new channel for core_cache(and any other cache entry), as well...

@flrgh Thank you!

@locao Done! It can be **REBASE** merged when all CI passed