Skip to main content

Alright. Since updating to JSS 9.3 (we've made each and every small jump since 9RC1) I've noticed (alright, I've been bashed over the head) Java's _appserver eating over 1000% CPU and running up inactive memory until the whole JSS locks up. Users utilizing "Self-Service" now cause a noticeable jump in CPU usage and enough RAM(inactive) so that if I don't watch it (think purge command) it will eventually lock up.



Running Casper Imaging will lock up the JSS within a few minuets. Imaging a dozen computers will lock up the machine (Showing full inactive memory and _appserver CPU utilization) within seconds.



Anyone else seeing this type of behavior?



P.S. here's my setup
JSS 9.3 and primary DP on
Mid 2010 MacPro w/8GB RAM running 10.8.5 and server 2.2.2
The Boot device happens to run the JSSs MySQL DB but that's a nice OWC SSD.
The Primary DP is running on an Arecra Internal RAID.



NetBoot and SUS on
Mid 2010 MacPro w/8GB RAM running 10.9.2 and server 3.1.1
Also, running a very fast Arecra Internal RAID for NetBoot and SUS/Caching Server



In any event, I've tried most quick things off the top of my head and am probably about to call my friends over at JAMF but, I figured that I'd throw this out there as well!

@galionschools Are you pushing the testing apps as In-House apps? Are they stored in the DB or are they on a web server?


@amanda.wulff, on my 9.3 test JSS, 10.9.2, Server 3.1.1, I tried your SMB suggestion for the DP... same results... imaged 1 machine and from 10GB... 3 minutes into the imaging/coping process, the available RAM went down to 1GB.



On a semi-related note, on my 8.73 production JSS, with no DP at all, 10.9.2, Server 3.1.1, and afp OFF... with 16GB of RAM, every morning the box shows like 5+GB of inactive RAM. I run sudo purge... the box behaves well for the rest of the day. The next morning all repeats again.



Just throwing it out there, the only thing I can think of of heavy taxing on the JSS box overnight are the backup and the machines reporting to the JSS...not sure then why the box shows like 5+GB of inactive RAM every morning?


Weird.. inactive RAM is memory that was used, is no longer being used, but is saved in case the application requests it again. It is essentially 'Free' memory.



Are you seeing low performance that gets better after the memory purge? If you don't do the purge, does the box not 'behave well'? Seems strange that it would be the Inactive Memory. Unless something has changed with Mavericks (I'm speaking of how the OS has worked up until ML).



thanks,



chris


Not in-house apps they're App Store apps. They aren't stored onsite either.


P.S. @amanda.wulff I just wanted to take a moment to thank you for all the work you've done with all of us on this. Amanda (along with a few others) was working with us on this issue the entire time on top of her posting here. I didn't put most of those conversations up because I, for some reason, didn't realize that you were the same Amanda that I was working with (Even though you said so). So, here's to you Amanda. You ROCK!



@lsmc08 OK, So now that I'm running both Casper related servers on 10.9.2, I've had several observations on RAM utilization and usage/reporting. First of all, are you experiencing any issues on the Mavericks servers even when they say that all Memory is used? Both of my boxes report using just about 100% of the memory, with almost no memory pressure. I have no idea what this means but my graphdat is showing my memory utilization at around 50%. I'm not sure how Apple really defines memory usage and memory pressure. I'm sure it's out there. I just wonder if it's finally doing what it's advertised, i.e. the ability to intelligently utilize all RAM without adverse performance.


@lsmc08 - Thanks for the update on that! I’ll jot it down in the notes I’ve got for this particular issue.



@Chris_Hafner - Thanks! I tend to use the JAMF Nation forums a lot for troubleshooting, as well as giving Technical Account Managers a heads up if I see one of their customers having an ongoing issue. A lot of times there’s something here that someone else has done or tried that I’ve either overlooked or wouldn’t have even thought to look at.


@Chris_Hafner



Alright... how about this. Is anyone who's experiencing this issue also seeing the following type of error in console
com.apple.launchd (com.jamfsoftware.jamfdsenroll[808]) exited with code 1

On top of that, ado you use JDS? I do not, yet I'm getting these JDS garbage errors. I'm looking for some form of corroboration here.


I'm seeing this on my test box but not on live. I just wiped out my test box and started over as I used a favorite app LaunchControl (http://www.soma-zone.com/LaunchControl/) to view how that launchd entry was configured. I discovered /usr/bin/jamfds, and also /Library/JDS. I do not use JDS.



It turns out the 9.25 JSS installer installed it by default. It is now gone on my test box.



chris


I like it. Fortunately it was solved for me by the time I ended up at 9.32. Thanks for the update though! It's going to be useful to many folks I'm sure!