| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
|
|
||
| ngx_http_upsync_server_t *upsync_server; | ||
|
|
||
| ngx_pool_t *pool; // worker common pool, can't be destroyed |
There was a problem hiding this comment.
更换为pool 的优点呢?有没有一种极端的情况,内存一直被吞噬(在增删server 比较频繁的场景下),pool 内存释放是以块为单位?
Sorry, something went wrong.
Sorry, something went wrong.
There was a problem hiding this comment.
欢迎任何可能的修改哈,探讨选取最优的;还有很多待优化的地方 :-);
@seven-eleven 谢谢
Sorry, something went wrong.
|
Thanks! The flag_array.size → flag_array.nalloc fix in this PR is already on master. Closing the rest: peers are allocated from a pool that is never destroyed, and ngx_pfree() only frees large pool allocations, so small allocations would never be released and memory would grow with every add/delete. |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
No description provided.