Bug 1421589

Summary:	there is no limit to storage wasted by multiple core dumps on migration destination
Product:	[oVirt] ovirt-engine	Reporter:	Denis Prezhevalsky <dprezhev>
Component:	BLL.Virt	Assignee:	Michal Skrivanek <michal.skrivanek>
Status:	CLOSED DUPLICATE	QA Contact:	meital avital <mavital>
Severity:	high	Docs Contact:
Priority:	high
Version:	4.1.0.4	CC:	ahadas, bugs, danken
Target Milestone:	ovirt-4.1.1	Flags:	rule-engine: ovirt-4.1+
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2017-02-21 10:28:41 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	Infra	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Denis Prezhevalsky 2017-02-13 08:16:50 UTC

Description of problem:
After upgrading engine & hosts and trying to migrate vm from not yet updated host, vm fails to migrate. Also, vm create multiple coredumps on each attempt to migrate, which in turn filled entire disk space on host (100%) and as result vdsm service failed.


Version-Release number of selected component (if applicable):
ovirt-engine-4.1.0.4-0.1.el7.noarch
vdsm-4.19.4-1.el7ev.x86_64

How reproducible:
None


Steps to Reproduce:
1.
2.
3.

Actual results:
Disk 100% full with coredumps, vdsm service failer, engine lost connection to host.

Expected results:
Show error/warning, limit coredumps size/number to not fill entire disk space.

Additional info:

Comment 2 Yaniv Kaul 2017-02-13 19:30:51 UTC

Not sure why it's virt - I believe host-deploy should fix and set it?
(also on RHVH - a separate bug?)

Comment 3 Michal Skrivanek 2017-02-20 14:39:37 UTC

there is core_dump_enable vdsm.conf parameter controlling this. I just thought it's disabled by default, but apparently it is not. 
Danken, do you recall more? 
I only know of the outstanding ABRT integration (bug 917062) which would change this, but that's not finished.
(since that other bug is infra I think the best owner is infra team)

Comment 4 Dan Kenigsberg 2017-02-21 09:55:20 UTC

core_dump_enable is on by default because qemu core dumps are rare and hard to reproduce. If users do not catch them at first crash, qemu-kvm team has nothing to work with.

But yes, proper ABRT integration would be much better than ovirt-specific tricks.

Comment 5 Michal Skrivanek 2017-02-21 10:28:41 UTC

ok, then this is intentional and RFE bug 917062 is tracking improvements in better debug information without a need for full memory dump

that particular crash is reported and tracked separately as bug 1421788

*** This bug has been marked as a duplicate of bug 917062 ***