Empty Extension attributes after upgrading to 11

[blochcw]

We successfully upgraded our on-prem environment to 11.0.1 last week and have noticed an odd issue with a number of machines showing no values for any extension attributes. Out of about 2000 current machines, it only appears to have affected about 180 or so. Manually running a recon on the machines has fixed most of them and I created a policy to just run recon as a process (not from maintenance) adn that seems to be slowly whittling them down and getting most of them back. Anyone else seen anything like this?

[sdagley · Accepted solution]

@blochcw Sounds like you may have run into PI113594:

"Performing multiple, overlapping inventory updates or API calls against the same computer record may result in the loss of extension attribute information on that inventory record. For example, a policy that includes an inventory update executed upon startup could conflict with a regularly occuring inventory update triggered by computer check-in."

[sdagley · Accepted solution]

@blochcw Sounds like you may have run into PI113594:

"Performing multiple, overlapping inventory updates or API calls against the same computer record may result in the loss of extension attribute information on that inventory record. For example, a policy that includes an inventory update executed upon startup could conflict with a regularly occuring inventory update triggered by computer check-in."

[blochcw]

OK, but is there a good solution for the issue? OR are we just waiting for the next patch?

[sdagley · Accepted solution]

Since it's an open PI I'm expecting Jamf will address it as there are definite uses for potentially overlapping recons, but for now it's kind of a "Patient: Doctor it hurts when I do X. Doctor: Then don't do X." situation.

[blochcw]

I have an open case and chatted with a support rep about. They are aware and working on it. In the meantime, I believe I found the source that was causing most of the issue for our implementation. I was using a webhooks server to pull inventory data from an Oracle db and posting it back Jamf. The trigger for the hook was on inventory. Fortunately, it was a redundent data source so I disabled the webhook and it looks like it may have fixed the issue. Definitely look for any redundent or repetitve API calls hanging around.