Aerospike cluster setup where multicast too powerfull

Operations Configuration
1

I have test servers and production servers. Now those are in same network and when I start serves those immediately go to same cluster which is nice but in my case how I can tell that test and production servers are separate clusters? I used this dun - Instruct this server to ignore another node (http://www.aerospike.com/docs/tools/asinfo/) …it worked, but I am not sure if I restart aerospike services or servers is this really so good to use. How i could tell that I have 3 servers in test ip1,ip2,ip3 and those should be test cluster and in produciton ip4,ip5,ip6 and those should be the production cluster.

And client should be able to access all (test, production) clusters.

thanks a lot=)! nice product to evaluate

I found this http://www.aerospike.com/docs/operations/configure/network/rack-aware/ should this used ? self-group-id ?

BR Ripa

2

Hi ripa,

Very nice to hear that you like the product. Using two multicast setups on the same network would not be recommended as you will be running into the very problem you are facing now. You could switch your TEST setup to A) a separate network or B) look at using MESH (unicast) for your TEST setup which would also have that effect. You can read more about MESH here: http://www.aerospike.com/docs/operations/configure/network/heartbeat/#mesh-unicast-heartbeat

Hope that helps!

Thanks Petter

3

It worked! I changed only for one server now, but mesh is working like it should →

[root@xxxx~]# asmonitor

Enter help for commands

1 hosts in cluster: xx.xx.xx.xxx:3000 Aerospike Interactive Shell, version 3.4.0

The only thing which was not so clear, page http://www.aerospike.com/docs/operations/configure/network/heartbeat/#mesh-unicast-heartbeat says mesh-seed-address-port 192.168.1.101 3002 but I used the old syntax,and the old syntax was fine. I already reinstalled 3.4.0 to see if config file syntax is changed but it wasn’t.

thanks for quick help!