Logic behind Distribution Point failover?

[donmontalvo]

So we have four Distribution Points. Each is set to fail over to another Distribution Point. Not that we'd ever do something like this, but the question came up, and we honestly didn't have an answer...

How does this work? I mean, yea, DP01 is down, so jamf goes to DP02. What happens if DP02 is also down? Does it go to its own failover which is DP03? Is it possible (please, not probable, just tossin' this out there) to end up with a loop?

Does jamf only try to failover once? Does it keep failing over? How does jamf manage these fail-overs? Are policies effected? Should we see the log say "Dang, DP01 is down, going to DP02, what, that's down too...hmm, let's try DP03...."

Just a thought, it would be nice to understand how jamf handles the actual failover when pushing policies.

Don

[myronjoffe · Accepted solution]

You could put the DP's behind a Load Balancer?

[kitzy · Accepted solution]

Don,

I was curious, so I just tried this in a test environment (v8.71) and it appears that it doesn't loop. It tries each server once and then returns an error back to the JSS.

[jdziat]

I have been told by JAMF support that it is possible to end up with a loop. So just make sure that you avoid linking to a previously used distribution point on fail. As far as logs go I believe it does mention it but to be honest its been awhile since I've seen it in one.

[myronjoffe · Accepted solution]

You could put the DP's behind a Load Balancer?

[kitzy · Accepted solution]

Don,

I was curious, so I just tried this in a test environment (v8.71) and it appears that it doesn't loop. It tries each server once and then returns an error back to the JSS.

[bentoms · Accepted solution]

AFAIK it only failsover once. Hence: https://jamfnation.jamfsoftware.com/featureRequest.html?id=46

[donmontalvo]

I passed this link to the group handling this. It's good to know a loop isn't possible. One fail over should be enough.