Load Balancer Health Check
We're trying to run moxi through our load balancer according to the 'gold standard'.
I configured our load balancer to do a GET /pools with HTTP basic authorization on each of our membase servers in the pool, however, the load balancer tells me that all our servers are 'UP' because membase is returning a status code 200 on the request, even when the server is a newly booted membase instance that has not joined any cluster yet.
What can I check for that will tell me if the server is actually a member of the cluster or not? I don't want moxi to send request to servers that aren't even in a cluster.
Thanks,
-Dave
Unfortunately, that also returns 200 OK.
:(
wget http://membase02:8091/pools/default
--16:01:15-- http://membase02:8091/pools/default
=> `default.1'
Resolving membase02... XX.XX.XX.XX
Connecting to membase02|XX.XX.XX.XX|:8091... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1,146 (1.1K) [application/json]
100%[========================================================>] 1,146 --.--K/s
16:01:16 (43.72 MB/s) - `default.1' saved [1146/1146]
-Dave
Dave,
Is the server that is not part of the cluster completely removed? (not just removed - pending rebalance)
What output do you get if you run the following:
[/opt/membase/1.6.x/bin/cli]#./membase server-list -c membase1:8091 -u user -p mypassword
and then
[root@redhat32 cli]# wget http://membase2:8091/pools/default (This should return 404 if the server is not returned in the output above; if you don't get 404, then its a bug and I will file in Jira)
Also, what version of membase are you running?
Thanks
Bhawana
1. It's not removed, it's never joined any cluster. A fresh install.
2. [dmangot@membase01 cli]$ ./membase server-list -c localhost:8091 -u membase -p secretpassword
ns_1@1.1.1.19 1.1.1.19:8091 healthy active
ns_1@1.1.1.20 1.1.1.20:8091 healthy active
[dmangot@membase01 cli]$ wget http://membase03:8091/pools/default
--2010-12-09 11:32:32-- http://membase03:8091/pools/default
Resolving membase03... 1.1.1.21
Connecting to membase03|1.1.1.21|:8091... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1146 (1.1K) [application/json]
Looks like a bug.
3. Name : membase-server Relocations: (not relocatable)
Version : 1.6.0.1 Vendor: (none)
Release : 1 Build Date: Wed 13 Oct 2010 01:58:18 AM PDT
Cheers,
-Dave
This looks like a bug that has already been resolved in the current build.
-Bhawana
Hey Dave, try changing your query to /pools/default. That should fail if a node isn't part of a cluster.
Perry
Forum support is great for free but sometimes you need a guaranteed response time and dedicated resources for your questions or issues.
Consider purchasing enterprise-level support from Couchbase: http://www.couchbase.com/products-and-services/overview
Call or email "sales -at- couchbase-dot- com" today!