Bug 1560146

Summary: anaconda invoked oom-killer: gfp_mask=0x14200ca(GFP_HIGHUSER_MOVABLE), nodemask=(null), order=0, oom_score_adj=0
Product: [Fedora] Fedora Reporter: Chris Murphy <bugzilla>
Component: kernelAssignee: Kernel Maintainer List <kernel-maint>
Status: CLOSED INSUFFICIENT_DATA QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 28CC: airlied, bskeggs, ewk, hdegoede, ichavero, itamar, jarodwilson, jglisse, john.j5live, jonathan, josef, kernel-maint, labbott, linville, mchehab, mjg59, steved
Target Milestone: ---Flags: jforbes: needinfo?
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2018-08-29 15:17:06 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
virsh dumpxml
none
journal log none

Description Chris Murphy 2018-03-24 02:42:50 UTC 
Description of problem:

I don't understand the oom kill well enough to assess if this behavior is expected, and if not whether it's a kernel or anaconda bug. But I read this:

https://www.kernel.org/doc/gorman/html/understand/understand016.html

And one thing it includes is:
 Is there enough swap space left (nr_swap_pages > 0) ? If yes, not OOM 

There was enough swap space left and yet oom killer was invoked. I think it's important to determine if the behavior is correct given 1.75G RAM and 2G swap, because getfedora.org only requires 1G RAM (which is probably too low)


Version-Release number of selected component (if applicable):
Fedora-Workstation-Live-x86_64-28-20180322.n.0.iso
4.16.0-0.rc4.git0.1.fc28.x86_64

How reproducible:
oom killer on something is fairly reproducible; what gets killed seems non-deterministic.


Steps to Reproduce:
1. Fedora 27 host 1792M RAM + 2G swap on /dev/vda
2. Boot Fedora-Workstation-Live-x86_64-28-20180322.n.0.iso
3. Launch the installer
4. Language selection, and test software disclosure is accepted

Actual results:

anaconda vanishes, and in the logs 
[ 2201.781664] localhost-live kernel: anaconda invoked oom-killer: gfp_mask=0x14200ca(GFP_HIGHUSER_MOVABLE), nodemask=(null), order=0, oom_score_adj=0


Expected results:

Either no crash; or up to date recommendations on minimum system requirements to avoid this type of problem.


Additional info:
Comment 1 Chris Murphy 2018-03-24 02:44:46 UTC 
Created attachment 1412384 [details]
virsh dumpxml

Details of the VM
Comment 2 Chris Murphy 2018-03-24 02:45:02 UTC 
Created attachment 1412385 [details]
journal log
Comment 3 Laura Abbott 2018-03-26 18:25:50 UTC 
Did this particular configuration work with other Fedora/kernel versions? That will be useful for reporting.
Comment 4 Chris Murphy 2018-03-27 02:01:25 UTC 
(In reply to Laura Abbott from comment #3)
> Did this particular configuration work with other Fedora/kernel versions?
> That will be useful for reporting.

It does work with Fedora 27 live media as released with even less memory (1.5G).

But I don't know when it started behaving as reported, so much has changed on live media all at the same time. A new thing I'm seeing in anaconda logs is /overlay (deleted) item as if maybe overlayfs is being used somewhere.

If I boot the live media and go directly to installation, it might work with 1.75G RAM, and it definitely will work with 2G RAM. But if I actually use the live media, like Firefox, or do some package updates which use memory on the dm overlay, all bets are off - the more the live environment is used the more unreliable it is. And also, it's not just anaconda that gets clobbered by the oom killer, I've seen gnome-shell get it, and packagekitd often gets hit by it.
Comment 5 Justin M. Forbes 2018-07-23 15:12:18 UTC 
*********** MASS BUG UPDATE **************

We apologize for the inconvenience.  There are a large number of bugs to go through and several of them have gone stale.  Due to this, we are doing a mass bug update across all of the Fedora 28 kernel bugs.

Fedora 28 has now been rebased to 4.17.7-200.fc28.  Please test this kernel update (or newer) and let us know if you issue has been resolved or if it is still present with the newer kernel.

If you experience different issues, please open a new bug report for those.
Comment 6 Justin M. Forbes 2018-08-29 15:17:06 UTC 
*********** MASS BUG UPDATE **************
This bug is being closed with INSUFFICIENT_DATA as there has not been a response in 5 weeks. If you are still experiencing this issue, please reopen and attach the relevant data from the latest kernel you are running and any data that might have been requested previously.