This service will be undergoing maintenance at 00:00 UTC, 2016-08-01. It is expected to last about 1 hours

Bug 511955

Summary: KVM based Guest Virtual Machine shut-down automatically
Product: [Fedora] Fedora Reporter: ali <aliahsan81>
Component: kvmAssignee: Glauber Costa <gcosta>
Status: CLOSED INSUFFICIENT_DATA QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: high Docs Contact:
Priority: low    
Version: 10CC: berrange, clalance, ehabkost, gcosta, markmc, mateenaslam, virt-maint
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2009-10-09 05:53:45 EDT Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:

Description ali 2009-07-15 15:18:09 EDT
Description of problem:
KVM based Guest Virtual Machine shut-down automatically

Version-Release number of selected component (if applicable):
kvm-74-10.fc10.x86_64

How reproducible:
It happen after 8 to 48 hours after running VM

Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info:
I got following in logs
Jul 12 05:01:43 huplo kernel: qemu-kvm[2651]: segfault at 284 ip 0000000000433bdf sp 00007fffbbaf4c70 error 4 in qemu-kvm[400000+193000]
Comment 1 Glauber Costa 2009-07-15 15:21:52 EDT
Please include at least a gdb backtrace.

you can attack gdb to the running pid of qemu-kvm, and whenever it crashes, do a 'bt'

otherwise it gets quite hard to even imagine what can be going on

thanks!
Comment 2 ali 2009-07-15 15:54:41 EDT
Well, This is a production box and have approx.10 Virtual Machines. At time the load is little high, KVM shut-down top VM and rest of machines got up and works normally. From logs i got the only error which I posted above. 

This simply means kvm dont crash completely. Plz let me know if still gdb backtrace works ?? then i can made available for you when i face the problem again on my box.


Regards
Comment 3 Glauber Costa 2009-07-20 11:28:05 EDT
backtracing in gdb should indeed work.
you are getting a SIGSEGV after all.

Also, please post the contents or /var/log/libvirt/qemu/<machine>.log
Comment 4 ali 2009-07-28 09:08:06 EDT
Right, after increasing RAM my VM is not shutting down now. I will submit the backtracing if i got the problem back on my box.
Comment 5 Mark McLoughlin 2009-08-07 10:58:12 EDT
ali: are you using an IDE or SCSI drive for the guest?

See also bug #512380
Comment 6 Mohammad Mateen 2009-10-02 07:09:26 EDT
I am getting the same problem 

Oct  2 10:02:08 kaloo kernel: qemu-kvm[3037]: segfault at 26c ip 0807dc67 sp bf9f2d90 error 4 in qemu-kvm (deleted)[8047000+175000]

I am using normal Sata. like https://bugzilla.redhat.com/show_bug.cgi?id=512380 i am getting nothing from bt as vm got shutdown and process terminats.

(gdb) bt
No stack.


Any info ? any fix ?
Comment 7 Mohammad Mateen 2009-10-02 08:57:49 EDT
Sorry i am using IDE with virtio driver.

----------------
02:00.0 IDE interface: Marvell Technology Group Ltd. 88SE6101 single-port PATA133 interface (rev b1)
----------------
Comment 8 Mohammad Mateen 2009-10-02 09:01:31 EDT
here are my log contents.

LC_ALL=C PATH=/sbin:/usr/sbin:/bin:/usr/bin /usr/bin/qemu-kvm -S -M pc -m 200 -smp 1 -name kaloo -nographic -monitor pty -boot c -drive file=/dev/vmdata/vm6.root,if=ide,index=0,boot=on -drive file=/dev/vmdata/vm6.var,if=ide,index=1 -drive file=/dev/vmdata/vm6.swap,if=ide,index=3 -net nic,macaddr=52:54:00:28:66:21,vlan=0,model=virtio -net tap,ifname=tap6,script=/etc/KVM/kaloo_tap6.sh,vlan=0 -serial pty -parallel none -usb 
TUNSETIFF: Device or resource busy
char device redirected to /dev/pts/12
char device redirected to /dev/pts/13
info cpus


Regards, Mateen
Comment 9 Mark McLoughlin 2009-10-09 05:53:45 EDT
mateen: this bug is against kvm in Fedora 10; please file a new bug for your issue

It looks like your problem is something wrong with your kaloo_tap6.sh script - the TUNSETIFF error is probably caused by the script

I'm closing this bug because the reporter isn't seeing the problem anymore