byteayan commented on PR #3355:
URL: https://github.com/apache/hugegraph/pull/3355#issuecomment-6437202450

   @imbajin thanks, all seven are in 826302f36. The paired doc PR is updated in 
apache/hugegraph-doc#535 (13737599).
   
   1. The auto budget is unchanged: heap, direct memory, the cgroup/JVM limit, 
the reserve and the shared Raft log `WriteBufferManager` all still apply.
   2. cgroup detection now lives in `CgroupMemoryLimit`. It returns the lowest 
visible limit, the cgroup version it came from, and whether the view is 
incomplete. `RocksdbMemoryBudget` only parses the setting and computes the 
budget. Nothing guesses at hidden limits: an incomplete view just logs a 
warning.
   3. The shipped config is `total_memory_size: auto`. A missing or blank 
value, or `0`, still means auto, and positive integers are bytes. 
`HG_STORE_ROCKSDB_TOTAL_MEMORY_SIZE=auto` passes the entrypoint, and there's no 
unit parsing.
   4. When `SPRING_APPLICATION_JSON` or `-D` sets the budget, 
`PdConfigureListener` drops only that key from the PD properties before adding 
the PD source. The synthetic override source is gone.
   5. `disruptorBufferSize()` and `writeBufferSize()` are back in `AppConfig`, 
with `AppConfigTuningTest`.
   6. `RaftRocksdbOptions.RAFT_LOG_CACHE_SIZE` builds the Raft log cache, and 
the budget reuses it as `RAFT_LOG_RESERVE`, plus a separate 512 MiB 
`NATIVE_RESERVE`.
   7. Startup logs one line, e.g. `RocksDB memory budget: 512 MiB (536870912 
bytes), mode=auto, physical-limit=4 GiB, source=cgroup-v2, heap=1 GiB, 
direct=512 MiB, reserve=1.5 GiB, cgroup-view=complete`.
   
   Store core suite: 162 tests pass, and the entrypoint test has 11 cases.
   


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to