Crowd : How best to ensure fault tolerance?

[Adam Flinton]

Dear All,

We are looking to use crowd as the central ID server. The one obvious thing about such a system is that it must be resilient. We are mostly cloud based (Digital Ocean) & as such can have many instances or possible instances of crowd.

So .. what is the best way to guard against machine/OS/app failure ?

Postgres replicating out?

Doing a nightly xml dump & importing it into clone crowd instances?

e.g. in essence if "crowd1" goes down for whatever reason we'd rather people/users didn't even know & as such "crowd2" takes over etc.

Has anyone done a multi-instance crowd setup with fault tolerance as a key goal?

TIA

Adam

[rrudnicki]

Hi Adam, 

There's a similar topic on this question where you also can find some ideias as well:

https://answers.atlassian.com/questions/10392005/best-practices-for-high-availability-for-atlassian-products

Also, there is this improvement request which I suggest you to vote:

https://jira.atlassian.com/browse/CWD-1053

To sum up, if you haven't already, we recommend you check out Go2Group's white paper on a clustering solution:http://www.go2group.com/en/services/atlassian-services/atlassian-crowd-a-simple-clustering-solution/. Whilst running Crowd in a cluster is unsupported, this may help some of you, although be aware that there are known unresolved configuration issues: https://jira.atlassian.com/browse/CWD-2925

Regards, 

Renato Rudnicki