SDK hangs when node in cluster fails over


#1

When I need to reboot a node in a 3 node cluster, and I use the failover option the node in the cluster, I notice that my SDK client hangs and does not fail or progress. Is there special configuration that needs to be done. Do I need to set the Observe Interval or Timeout?


#2

@envitraux -

Which client version are you using?

-Jeff


#3

From NuGet 2.0.0.1

Here is the config

_config = new Couchbase.Configuration.Client.ClientConfiguration
{
Servers = new List {
new Uri(“http://bnc-cb2.xxx.corp:11211/pools”),
new Uri(“http://bnc-cb1.xxx.corp:11211/pools”),
new Uri(“http://bnc-cb3.xxx.corp:11211/pools”),
},
BucketConfigs = new Dictionary<string, Couchbase.Configuration.Client.BucketConfiguration>
{
{ “mandabase”, new Couchbase.Configuration.Client.BucketConfiguration
{
BucketName = “mandabase”,
UseSsl = false,
Password = “xxxx”,
PoolConfiguration = new Couchbase.Configuration.Client.PoolConfiguration
{
MaxSize = 10,
MinSize = 5
}
}
},
{ “mandabase-xxxx”, new Couchbase.Configuration.Client.BucketConfiguration
{
BucketName = “mandabase-xxxx”,
UseSsl = false,
Password = “xxxx”,
PoolConfiguration = new Couchbase.Configuration.Client.PoolConfiguration
{
MaxSize = 10,
MinSize = 5
}
}
}
}

        };

#4

@envitraux

Update to 2.0.3 (it’s on Nuget now: https://www.nuget.org/packages/CouchbaseNetClient/2.0.3) - there has been a lot of internal changes since 2.0.0.1.

We’ll take it from there?

-Jeff


#5

I will try it, and see if it helps


#6

I think that worked. Only could test once.


#7

I may have a similar problem.
1st experiment:
I am using server 3.0.1 Community Edition (build-1444)., sdk 2.0.1.
I added third server to the cluster and triggerd rebalance. After some time my application stopped responding. The cpu of iis proces went to 100% and all RAM was used. I found this thread so i tought that i would update sdk library.
What i did today with 2.0.3 sdk: I failed over one server. It went ok, all working fine. I restarted the server and after it came back i clicked full recovery. Then i clicked rebalance. I got the same situation as yesterday. 100% cpu and some kind of memory leak.


#8

@drak25 -

Could you tell me about your deployment environment? For example was it ASP.NET/IIS?

Also, would you mind opening a Jira ticket for further investigation? You can do so here: https://issues.couchbase.com/browse/NCBC

Thanks!

-Jeff


#9

I took a crash dump while the problem was occuring and passed it through windbg. I realized that the problem is probably
https://github.com/couchbaselabs/couchbase-aspnet component. The version that i am using references couchbase sdk 1.3.9 and it may cause problems. I will update it to use 2.0.3 and check if the problem still occurs.


#10

@drak25 -

The Couchbase ASP.NET Session and Cache providers has an updated version: https://www.nuget.org/packages/CouchbaseAspNet/2.0.0-beta

It’s been updated to v2.0.3 of the SDK and requires some tweaks to make it work (see the link in the release notes section). Sometime later this year we hope to release it formally, as it is still a beta and not fully tested.

The project is on github: https://github.com/couchbaselabs/couchbase-aspnet/tree/2.0

If you feel inclined to contribute, just push a pull request :smile:

-Jeff


#11

Thank you but for now i just took the old code and tweaked it to use 2.0.3.

I have a quick question btw. Why when you check the version of assembly of 2.0.3 it says 2.0.2.0 ?


#12

@drak25 -

That looks like a packaging bug…I’ll push an updated package with the correct version.