In the background work, Kafka has been used all the time, and several Kafka clusters are also maintained. There is no doubt that Kafka is the most popular message queue middleware at present, and what makes it popular

PM2 log cutting requires the installation of the pm2-logrotate plug-in. The plug-in configures logs for automatic log cutting, and automatically deletes expired logs to prevent logs from filling the directory. Installation steps

The high availability and self balancing scheme of the ElasticSearch cluster will automatically copy the partition of the node on other nodes after the node hangs up (restarts), which will result in a large amount of IO overhead. If you leave

Why do you want to monitor the segments of the online es cluster of Segments Force Merge? See the following figure: If you cannot understand the segment, you can view it

Direct figure above: The index of indexElasticsearch is similar to the library in a relational database. Index is the general name of the unit that stores data. But actually, this is just Cheng

The colleague of the company wrote an article on the configuration of the CPU performance mode, which was very good, so he moved it directly to record it. With the ib network card, the CPU can get higher response speed in high-performance mode after starting the connection by default

During normal development, testing, and even production environments, OutOfMemoryError is sometimes encountered, and the Java heap overflows, which indicates that there is a serious problem with the program. We need to find the cause

Heap DumpA Heap Dump is a snapshot of the memory used by a Java process at a certain time. That is, at a certain time, the memory of the Java process

1. Create topic/ kafka-topics.sh --create --zookeeper ip:port --replication-f...

Here's the thing: the replica policy has been adjusted: 1. The es cluster has many machines and nodes. In order to achieve higher stability, it is decided to enable the storage policy of one primary partition and two secondary partitions, that is, one copy of data is stored in three

Because es only has access permission restrictions, it does not do operation permission verification. To do operation permission verification, you need to spend money to buy an X-pack license, so the data operation in es is random

The Elasticsearch index defaults to one shard replica. If there are more nodes, the more likely that two nodes will fail at the same time, then you need to reconfigure the number of shard replicas