I have 4 buckets, of which only 1 bucket has 1 item. Why does re-balancing take so long when adding an additional server?
The rebalance could take some time depending of the speed of your infrastructure (IO,network, CPU) and the tupe of data (views or not, ...)
What surprised me is the fact that you are saying that you have one single bucket with 1 item, so could you tell us:
- what is your hardware? on each node
(memory, disk,.. type of network between node)
- Operating system?
- what do you call "long rebalance" ? (how long it takes to do the rebalance?
- Do you have any activity of the cluster when doing the rebalance?
- No sample buckets? views?
This cluster is made up of 2 nodes. For some reason, the first node created always shows high CPU, mostly coming from the erl.exe. This probably has something to do with the re-balancing timing. What is this .exe doing that it requires so much CPU?
CPU: Intel Xeon X5680 @ 3.33GHz
RAM: 4 GB
System type: 64-bit operating system running Windows Server 2008 R2 Standard
Not sure what is meant by "type of network between node". These servers are in our DMZ.
When I say "long rebalance", this equates to over an hour at least.
No other significant activity on the cluster while doing the rebalance.
No sample buckets, just 4 buckets I created.
Let me know if you need additional info.
© 2013 COUCHBASE All rights reserved.