Binbin
yean, active-rehash-percent seems better, sorry for the confusing (bad English aha). yes, we don't need it now i guess. It is a option in case this one got merged (fixed...
@valkey-io/core-team please take a look with this comment https://github.com/valkey-io/valkey/pull/977#issuecomment-2323800261 and see if we want to handle more the other one is this PR (and its top comment), not sure if...
@gmbnomis Thank you for your summary comment @valkey-io/core-team Please also take a look at the comments above and vote.
> 5. zfree uses bio thread when dict exceeds a certain size, such as d->ht_used[0] > 4096 i see it was dismissed, is this still valuable? I know that in...
ohh, in fact, my previous test was done on an old version (redis-4). I verified that when jemalloc-bg-thread yes is enabled, the max latency can be avoided.
> Looking at the code for incrementallyRehash, I can see that if some dict was rehashed, then it does no active rehash work for the expires dict nor for other...
@oranagra btw, out of the topic, do we have any idea to include rehash dict overhead in freeMemoryGetNotCountedMemory? In my fork, we included it assuming it is a temporary memory...
> An idea for 4. If we aren't spending any time sleeping in the event loop, it means the load is high, right? Can we skip incremental rehash if we...
> If we do active rehashing only if there is no traffic at all (there are no file events between time events), then if there is just one I/O event...
It was approved by @zuiderkwast back in https://github.com/redis/redis/pull/13051