FYI, here are the debug logs as well. Don't see any errors as such. But, there are several warnings.
Is there a minimum VM type I should be choosing, I deployed this on m4.large.
Should that be the issue?
EDIT: I did try running a large VM type, the behavior is still the same as reported above.
VM size here won't make a difference, this only really impacts performance based on your graph size.
The login screenshot shows that neo4j is running and accepting connections (although you used the wrong password). Presumably you can log in using the correct password? This means that the broken web page on browser is more likely to be about some network setting in between.
Because we needed to get you a new image, you must have had to make changes to the cloudformation because as discussed above you didn't have 3 AZs in your region, correct? I would look back over those changes to see what else might have happened.
Oops, yes I can login to the shell with correct password.
Well, in this case I actually ran it in ap-southeast-1 (Singapore) region without any change to the template directly from the listing: AWS Marketplace: Neo4j Enterprise Causal Cluster
And, I'm unable to access via the < public-ip >:7473 via on neither of the 3 EC2's. But, I can ssh and login into the shell.
I'll have to make changes to the template for my needed zone, but need to figure out what exact changes would be needed for that. If any pointers on this will be great too?
But, even if I can have the existing template work on ap-southeast-1 should work for now.
I will try to reproduce this issue in ap-southeast-1 from the marketplace listing and investigate later today.
I've just tried this, and the deploy went fine from the marketplace in ap-southeast-1, and I was able to connect from the outside with my browser, so I can't seem to reproduce the issue.
Can you check your CloudFormation resources and see if there are any errors or warnings?
Another thing you might check is the /etc/neo4j/neo4j.conf file on any of the three machines, make sure it has an advertised_address that matches the external IP of the machine.
Got it.
I have access too. But, port 7473 on any of the public ip isn't accessible. Only the one listed in the Output tab is:
In my case it was the Node1's ip, I was trying to access via Node0's ip.
Works for now, thanks again for your assistance.
Will be trying to work on the template for my region, but I think I will go with ap-southeast-1 for the moment.
Arnab
FYI,
If anyone is interested in deploying this in ap-south-1 (Mumbai) region, I've reworked the CloudFormation template to be able to so with ami-0284a2c822c6c3b9e
https://s3.ap-south-1.amazonaws.com/cf-templates-1wqj6b0ycugib-ap-south-1/neovm-2az.yml
Could you describe quickly what were the major changes you made?
Sure.
The major change was to remove 'Subnet 1' as there are only two availability zones in this region. So, I moved the resources under Subnet 1 i.e One core server, and two read replicas into Subnet 0, ensuring all the other routing stays intact.
This helped to achieve the same 3 core VM but with 2 subnets instead of 3. Performance wise I'm yet to experience if there is any issues. But, since, this is closest to my application users the latency would definitely be the least.
