Skip to main content

High availability with Glassfsh, some open questions

1 reply [Last post]
tomar99
Offline
Joined: 2010-05-12
Points: 0

Hi,

I am currently working with Glassfish, with a single domain. (1 application)
After few outages on my application, I started to look at installing several glassfish instances and put load balancing there (with apache and mod_jk).
I followed the oracle tutorial, and the load balancing is working well so far.

My questions, so far:
- Apache: How to avoid single point of failure ?
- Glassfish:
- Is clustering the only possibility to set-up load balancing?
- If I'm not mistaking, there is only one domain, administrating and managing the several instances, right? What happens if the node where the domain is running crashes? Everything is down? How to avoid that? Passive domains on the other nodes?
- Is there other point to consolidate?

I am quite new in the glassfish world, and totally new with high availability challenges.
Please do not hesitate to share comments with me, podcasts, documents, good practices to have a highly available architecture.

Thank you in advance !
Romain.

Reply viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
josesuero
Offline
Joined: 2009-07-17
Points: 0

I'm no glassfish expert or with oracle, but I'll share my experience on clustering

- Apache: How to avoid single point of failure ?
We currently have two load balancers with round robin DNS

- Glassfish:
- Is clustering the only possibility to set-up load balancing?
It really depends on your application clustering helps distribute your app on instances and share session information, if you don't need session sharing you can have several instances of glassfish on different servers and that should work as well.
- If I'm not mistaking, there is only one domain, administrating and managing the several instances, right? What happens if the node where the domain is running crashes? Everything is down? How to avoid that? Passive domains on the other nodes?
if the node manager fails instances will still be running since they don't depend on the DAS (Domain administration server) to be running
- Is there other point to consolidate?
Don't know what you mean