"Reset Quota" after failover


#1

What is the purpose of the “Reset Quota” button after a failover has occurred? There doesn’t seem to be any documentation on it. Version is Community 2.2.

I’m not able to post an image but the dialog is “A server was automatically failed over” and “Reset Quota”


#2

We only allow AUTO failover of one node within a cluster at a time. This is to prevent the thundering herd scenario. When a node is AUTO failed over, the quota must be reset to allow the cluster to auto failover another node (assuming the replica count is correct and current)


#3

Hi,

By default this quota is to 1. Do you know how to change it ?

Thank you


#4

@Crafty33 there is no easily accessible way to change the quota to something greater than 1.


#5

How about the hard way?
Couchbase failure scenarios tend to encourage downtime, which is a real problem in the minds of system administrators accustomed to memcached (where a failure of a node doesn’t cause an outage for a whole farm.) When a couchbase node fails, the cluster freezes until a failed node is failed over. This is hard for me to explain in a way that’s acceptable to people who wear the pager for an environment that really tolerates zero downtime, particularly when the point of failure is just a cache cluster. It’s especially hard to explain because Couchbase is supposed to be all about high availability and resilience.
Has anyone developed a monitor system that can make a Couchbase cluster more reliable from the point of view of a large number of clients? The scenario we discuss is “taking a Sawzall to two drives shouldn’t cause a total outage of the farm.”