I'm currently working on simle active stand/by Redis clustering for query caching with Keepalived.
The problem I'm facing is when I shutdown&start(restart) the host1 server for testing, it should keep the exact same virtual ip, and start replication with host2. But syslog outputs "Unable to connect to MASTER: Resource temporarily unavailable : Connecting to MASTER no:0", and fail the replication."
I assumed that would be happened because of OOM, and the kernel might kill the process. So I implement these below refering the Redis set-up hints (https://www.google.co.jp/search?q=redis+administrator&oq=redis+administrator+&aqs=chrome..69i57j69i64.4847j0j1&sourceid=chrome&es_sm=91&ie=UTF-8).
- vm.overcommit _memory = 1
- set up swap file
- made sure that I have enough amount of memory for redis ( 2 times the memory usually used )
Do you have any ideas? Thanks,