docs: linux, raft, flink

content/posts/Raft-Consensus-Algorithm.md

@@ -0,0 +1,12 @@
++++
+title = "Raft共识协议的一些细节"
+date = 2022-10-13
+[taxonomies]
+tags=["Raft","分布式"]
++++
+## Log compact机制
+snapshot中需要对log进行compact，防止log过多对服务造成大量压力。
+### compact策略
+- 执行snapshot时，删除上次snapshot之前的数据
+- 执行snapshot时，固定保留最后n条entry
+- leader计算近期（分钟级别）心跳中所有follower的next_index最小值，再保留前n条entry

content/posts/flink-kubernetes-standalone.md

@@ -0,0 +1,14 @@
++++
+title = "Flink standalone on kubernetes"
+date = 2022-09-08
+draft = true
+[taxonomies]
+tags=["flink","document"]
++++
+
+## start job
+### start job from savepoint on flink webui
+- select job jar
+- input job class refrence path
+- configuration global `parallelism`, `args`, `savepoint path`
+> eg: use s3 savepoint dir `s3://flink/savepoint/savepoint-ee796f-2850304e783a`

content/posts/linux-command.md

@@ -0,0 +1,10 @@
++++
+title = "一些linux命令"
+date = 2022-09-15
+[taxonomies]
+tags=["linux"]
++++
+### 带进度的wc
+```bash
+awk 'BEGIN {T=0} (T!=systime()) { printf "%s %s\n",NR,$0 ; T=systime()} END { print NR}'
+```