pbchou commented on issue #8728: URL: https://github.com/apache/trafficserver/issues/8728#issuecomment-1069500337
It sounds like your use case is different than ours. This might be something that could be included as a plugin parameter to decide whether to run the garbage collector on remap instance deletes. Let me explain why we added the garbage collector calls. Our current production config generates close to 1GB of Lua memory usage. With the current LuaJIT v2.1.0-beta3 that is standard, the memory usage is limited to 2GB on Linux. This means that we could not reload remap config more than once before running the garbage collector. If one is willing to take the latest non-tagged LuaJIT snapshot, the 2GB limit is gone. However, then you would eventually run into system memory limitations. We are very concerned about memory usage as we would prefer to use the RAM for caching. Is there a way to set and guarantee a maximum amount of time or memory usage threshold between garbage collector cycles? Otherwise, how would we engineer our network controller in terms of pushing config files and executing config reloads? Out of hundreds of caches maybe some have run GC and some have not. FYI, in looking through the LuaJIT source, there is a garbage collector option to stop garbage collection. It sounds like this might be useful to you since Lua deciding on its own to run the GC in your use case might be a time bomb. ``` lua_gc(L, LUA_GCSTOP, 0); ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
