This is an automated email from the ASF dual-hosted git repository.
jiacai2050 pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/horaedb-docs.git
The following commit(s) were added to refs/heads/main by this push:
new 9dcbfd4 feat: Add user guide for compaction offload (#158)
9dcbfd4 is described below
commit 9dcbfd41da3b6bee6dd161ff2764f1e705669640
Author: Leslie Su <[email protected]>
AuthorDate: Fri Nov 22 14:56:14 2024 +0800
feat: Add user guide for compaction offload (#158)
Co-authored-by: Jiacai Liu <[email protected]>
---
.../docs/user-guide/cluster_deployment/with_meta.md | 21 ++++++++++++++++++++-
1 file changed, 20 insertions(+), 1 deletion(-)
diff --git a/content/en/docs/user-guide/cluster_deployment/with_meta.md
b/content/en/docs/user-guide/cluster_deployment/with_meta.md
index 1141a46..1023604 100644
--- a/content/en/docs/user-guide/cluster_deployment/with_meta.md
+++ b/content/en/docs/user-guide/cluster_deployment/with_meta.md
@@ -187,9 +187,28 @@ timeout = "5s"
server_addrs = ['http://{HoraeMetaAddr}:2379']
```
+### Compaction Offload
+
+Compaction offload is also supported. To enable compaction offload, the
corresponding compaction mode with node picker and endpoint should be
configured.
+
+- `node_picker`: There are two types of node picker -- `Local` and
`Remote`(WIP).
+ - When the `Local` is setted, the local compaction task would be offloaded
to the specific remote compaction server, which decided by `endpoint`.
+- `endpoint`: The endpoint, in the form `addr:port`, indicating the _grpc
port_ of the remote compaction server.
+
+Here is an example for it:
+
+```toml
+[analytic.compaction_mode]
+compaction_mode = "Offload"
+node_picker = "Local"
+endpoint = "{RemoteCompactionServerAddr}:{RemoteCompactionServerGrpcPort}"
+```
+
+A Compaction Server, responsible for executing the compaction task, is also
needed. Currently `horaedb-server` will act as this role, in the future we can
move it to an independent service.
+
### Complete Config of HoraeDB
-With all the parts of the configurations mentioned above, a runnable complete
config for HoraeDB can be made. In order to make the HoraeDB cluster runnable,
we can decide to adopt RocksDB-based WAL and local-disk-based Object Storage:
+With all the parts of the configurations mentioned above, a runnable complete
config for HoraeDB can be made. In order to make the HoraeDB cluster runnable,
we can decide to adopt RocksDB-based WAL and local-disk-based Object Storage
without compaction offload:
```toml
[server]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
