Upgrade to 10.3 followup

[andrew_nicholas]

We just took the plunge to Jamf Pro 10.3 to help address the KEXT non-sense, and while things in test were fine the plunge to prod has been mixed. The JSS seems more sluggish and tomcat is a much bigger memory hog than it was previously, macOS MDM enrollment and DEP completion fail because of timeouts, and on top of that the DB seems extra prone to fault with back up failures repeatedly. We had some issues with the box that could have lent a hand to all of this but those were seemingly resolved before we went forward with the upgrade. Has anyone else noticed this behavior afterwards and/or is there any server side tweaking that seems to have helped?

[millersc]

Just curious, which platform is the server running on?

[thoule]

10.3 is running slightly faster for us than 9.x. As for memory, it's aways been a hog, so no change there. I'd suspect some DB issues and recommend contacting support. Or perhaps your mysql database doesn't have enough RAM.

[m_donovan]

What does your environment look like? I am planning a JamfPro 10 upgrade from 9.101 this Thursday.

[dgreening]

One of these days Jamf will get around to documenting the InnoDB conversion process, which should have become standard YEARS ago.

[ejculpepper]

We just upgraded our environment from 9.101 to 10.3, it seems like our Jamf console is responding much quicker then it did on 9.101. I haven't noticed a spike in memory usage on our servers either.

The issues you're describing sound like issues we've seen when our DB was having problems. We've had issues where SQL was holding onto threads from our Jamf servers, causing there to be no open threads for new requests which then caused failed enrollments, devices not responding to management commands, and sluggish performance.

[andrew_nicholas]

@millersc we run on Server 2012r2 but from what others are saying I'm going to focus on the DB being the likely source. Thanks all for the feedback!

[cbrewer]

@andrew.nicholas We made the jump from 9.101 to 10.3 this weekend. The upgrade itself was painless. The experience afterward, however, was similar to what you're reporting. In my case, it was bad enough that I rolled back to 9.101. I am hoping to work more with Jamf support this week and identify possible causes. 9.101 had been running great for months. We're already on InnoDB and performance issues have been near nonexistent lately. I'm still hopeful that my 10.3 issues were something unique to my environment, but I also have concerns that version 10 may just need more time.

[CGundersen]

Large clustered environment w/ many sites still running v9 here. We are already running InnoDB (VxRail) and performance isn't great for operations (e.g. configuration profile) across sites (e.g. all devices). We won't be able to tolerate a performance hit here. Watching this thread ...

[dgreening]

I'll say that for a global environment with 25k+ managed Macs, 10.3 on an InnoDB converted DB has been working quite nicely. Once hosting gets our DB host VM on even faster storage we will be sitting even nicer. The saved search / smart group re-factoring in 10.3 sure helps. We DO have to routinely to some manual DB work to keep our log flushing working (purge orphaned computers from the reports table), cutting down the locations and location_history tables, etc.. There is still quite a bit of manual MySQL work you need to do to keep your DB lean and mean...

[m_donovan]

@dgreening Are you using user initiated enrollment in your environment. That is the only concern my STAM had with our planned upgrade coming up. Also would you be willing to share the orphaned computer cleanup process you use. I periodically clean up locations and osx config profiles but don't have a workflow for orphaned computers.

[dgreening]

@m.donovan I need to write it up for sure. You can also get the cleanup steps from your STAM (thats where I got mine).

[cbrewer]

@andrew.nicholas How is 10.3 working now that it's been a couple weeks?

@dgreening Are you using memcached? Also, how many iOS devices?