Automatic fail-over


#1

We have Couchbase 4.0.0 linux servers with data partitions mounted with errors=remount-ro option:

UUID=some-hex-appears-here /couch ext4 errors=remount-ro 0 1

When filesystem error occurred our data partition has been forcibly
remounted to read-only mode. In this case we expect that the couchbase
instance on the server switch into Pending state. But this switching is
not happening. The server continue work in
Online state.

How to reproduce:

A command
echo u > /proc/sysrq-trigger
forcibly switch server’s file systems to read-only mode.