Insufficient consumption performance leads to a large number of blocking situations in Redis. #1897
Description
The current status of my ES cluster：
12c16G x 3 version：5.6.9
Redis：
4c4g version：7.2.1
Among these four jobs, different Redis databases were used. For instance, the first job utilized db1, the second used db2, and so on. The figure shows the queue situation of one of the databases. In fact, this is the case for all four databases.
Queue situation
Exceptionless：
A total of 4 jobs are running on a 4-core 32GB Windows server.
This problem has been bothering me for a week. Even though Redis was upgraded from 2c4g to 4c4g, it still didn't work. Similarly, Elasticsearch was upgraded from 4c8G to its current state, but it also failed. When attempting to run 7 instances of the job, there was still a significant write delay, with the delay being around 1 hour. Not all logs had the delay, but some log queues were blocked. This was because in Exceptionless, querying logs always returned the latest <= 1 second, but it was incomplete because some were still blocked.