Hash Set Key and fields expiration

[ltagliamonte-dd]

Hello Folks,
I'm trying to understand better kvrocks internals.

In the specific I'm trying to understand what part of the codebase take care of the logic behind compaction the hash keys.
The number of keys in my cluster is steadily growing so I'm looking into 2 directions:

• how many new keys the app is inserting
• if there is any leak in the compaction implementation in kvrocks

I want to make sure that when I have a TTL on an Hash Key, during compaction the TTL on the key is evaluated and all it sub fields are correctly removed.

Can someone please point me to the logic in the codebase?

Thank in advance for all the help!

[git-hulk]

how many new keys the app is inserting

The one impossible way is to use DBSIZE SCAN to collect the key count, but this won't be precise since the keys might expire.

if there is any leak in the compaction implementation in kvrocks

I think we cannot know if any fields are leaked after the compaction, but we can add basic C++ unit test cases to promise this:

1. Add hash with part of fields with short expiration time
2. Sleep to wait for fields to expire and trigger compaction
3. Scan fields with the hash key prefix to confirm if expired fields were disappeared

[ltagliamonte-dd]

Thank you for the answer @git-hulk.
Can you please point me to the code so i can take a look into adding an unit test?
Would an integration test in golang work as well?