[incubator-doris] branch master updated: [doc] The description of the metadata design document is not clear (#9367)

[mo...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

morningman pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-doris.git


The following commit(s) were added to refs/heads/master by this push:
     new 2970f9842e [doc] The description of the metadata design document is not clear (#9367)
2970f9842e is described below

commit 2970f9842e969140a06ffcea6d44dfc893481e1d
Author: qimumu9406 <10...@users.noreply.github.com>
AuthorDate: Thu May 5 20:48:32 2022 +0800

    [doc] The description of the metadata design document is not clear (#9367)
---
 docs/en/design/metadata-design.md    | 2 +-
 docs/zh-CN/design/metadata-design.md | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/docs/en/design/metadata-design.md b/docs/en/design/metadata-design.md
index 43a2fa89b1..d291c8ef6e 100644
--- a/docs/en/design/metadata-design.md
+++ b/docs/en/design/metadata-design.md
@@ -73,7 +73,7 @@ The data flow of metadata is as follows:
 
 2. After the log is written to bdbje, bdbje copies the log to other non-leader FE nodes according to the policy (write most/write all). The non-leader FE node modifies its metadata memory image by playback of the log, and completes the synchronization with the metadata of the leader node.
 
-3. When the number of log bars of the leader node reaches the threshold (default 10W bars), the checkpoint thread is started. Checkpoint reads existing image files and subsequent logs and replays a new mirror copy of metadata in memory. The copy is then written to disk to form a new image. The reason for this is to regenerate a mirror copy instead of writing an existing image to an image, mainly considering that the write operation will be blocked during writing the image plus read lock. [...]
+3. The number of log entries of the leader node reaches the threshold (default 10w) and meets the checkpoint thread execution cycle (default 60 seconds). Checkpoint reads existing image files and subsequent logs and replays a new mirror copy of metadata in memory. The copy is then written to disk to form a new image. The reason for this is to regenerate a mirror copy instead of writing an existing image to an image, mainly considering that the write operation will be blocked during writi [...]
 
 4. After the image file is generated, the leader node notifies other non-leader nodes that a new image has been generated. Non-leader actively pulls the latest image files through HTTP to replace the old local files.
 
diff --git a/docs/zh-CN/design/metadata-design.md b/docs/zh-CN/design/metadata-design.md
index a1cdd29ee2..58e2772c4a 100644
--- a/docs/zh-CN/design/metadata-design.md
+++ b/docs/zh-CN/design/metadata-design.md
@@ -72,7 +72,7 @@ Doris 的元数据是全内存的。每个 FE 内存中，都维护一个完整
 
 2. 日志写入 bdbje 后，bdbje 会根据策略（写多数/全写），将日志复制到其他 non-leader 的 FE 节点。non-leader FE 节点通过对日志回放，修改自身的元数据内存镜像，完成与 leader 节点的元数据同步。
 
-3. leader 节点的日志条数达到阈值后（默认 10w 条），会启动 checkpoint 线程。checkpoint 会读取已有的 image 文件，和其之后的日志，重新在内存中回放出一份新的元数据镜像副本。然后将该副本写入到磁盘，形成一个新的 image。之所以是重新生成一份镜像副本，而不是将已有镜像写成 image，主要是考虑写 image 加读锁期间，会阻塞写操作。所以每次 checkpoint 会占用双倍内存空间。
+3. leader 节点的日志条数达到阈值（默认 10w 条）并且满足checkpoint线程执行周期（默认六十秒）。checkpoint 会读取已有的 image 文件，和其之后的日志，重新在内存中回放出一份新的元数据镜像副本。然后将该副本写入到磁盘，形成一个新的 image。之所以是重新生成一份镜像副本，而不是将已有镜像写成 image，主要是考虑写 image 加读锁期间，会阻塞写操作。所以每次 checkpoint 会占用双倍内存空间。
 
 4. image 文件生成后，leader 节点会通知其他 non-leader 节点新的 image 已生成。non-leader 主动通过 http 拉取最新的 image 文件，来更换本地的旧文件。
 


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org