configuration of ignite client nodes

classic Classic list List threaded Threaded
2 messages Options
Scott Cote Scott Cote
Reply | Threaded
Open this post in threaded view
|

configuration of ignite client nodes

We have ignite client nodes (2.7.0 linux)  embedded inside our microservices (spring boot apps) pointed at a dedicated ignite server (2.7.0 linux).  Each client node has its ip finder – the vm ip finder –configured to seek the dedicated server.

 

None of the client nodes are configured to know about each other.

 

Should the client nodes know about each other?  In other words, do I need to place the ip address of the client nodes in the vm ip finder list for each of the clients?

 

We have the apps (Abbreviated names and abbreviated addresses):

 

QM – 19

ALPR – 18

RM – 21

A – 20

 

Ignite Server – 225

 

QM, ALPR, RM, and A all have their VM ip finder list to only seek “225” .   Do they need to have each other in there too?   We keep seeing the client nodes trying to seek each other out inside the spring boot logs.

 

Another note:

Using Visor ping command – usually at least one node fails in the ping.  Sometimes they all pass, but not often.  How do I leverage this diagnostic?

Using Visor cache -scan sometimes fails …. Symptom again?

 

Have increased clientFailureDetectionTimeout from 30,000 to 100,000 on the server.

Saw the message to increase networkTimeout from 5000.  So increased it to 20,000 on the server.   Do we need to do this on the client too?

 

What is a symptom of problems from mixed ip environments?   See that warning in the log also…..

 

SCott

stephendarlington stephendarlington
Reply | Threaded
Open this post in threaded view
|

Re: configuration of ignite client nodes

The list of machines in your IP finder list does not need to be exhaustive. As long as a node can find at least one other it should be able to join the cluster.

You don’t need to configure your clients to know about the other client nodes, but, by virtue of joining the cluster, they will learn about all the other nodes, both clients and servers.

Thick clients are full participants of the cluster, so the same timeouts on the clients as well as the servers would make sense.

Regards,
Stephen

On 8 Jul 2019, at 04:22, Scott Cote <[hidden email]> wrote:

We have ignite client nodes (2.7.0 linux)  embedded inside our microservices (spring boot apps) pointed at a dedicated ignite server (2.7.0 linux).  Each client node has its ip finder – the vm ip finder –configured to seek the dedicated server.
 
None of the client nodes are configured to know about each other.
 
Should the client nodes know about each other?  In other words, do I need to place the ip address of the client nodes in the vm ip finder list for each of the clients?
 
We have the apps (Abbreviated names and abbreviated addresses):
 
QM – 19
ALPR – 18
RM – 21
A – 20
 
Ignite Server – 225
 
QM, ALPR, RM, and A all have their VM ip finder list to only seek “225” .   Do they need to have each other in there too?   We keep seeing the client nodes trying to seek each other out inside the spring boot logs.
 
Another note:
Using Visor ping command – usually at least one node fails in the ping.  Sometimes they all pass, but not often.  How do I leverage this diagnostic?
Using Visor cache -scan sometimes fails …. Symptom again?
 
Have increased clientFailureDetectionTimeout from 30,000 to 100,000 on the server.
Saw the message to increase networkTimeout from 5000.  So increased it to 20,000 on the server.   Do we need to do this on the client too?
 
What is a symptom of problems from mixed ip environments?   See that warning in the log also…..
 
SCott