Bug 601619 - migration never ends with default migrate_downtime under high network stress
Summary: migration never ends with default migrate_downtime under high network stress
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Enterprise Linux 6
Classification: Red Hat
Component: qemu-kvm
Version: 6.0
Hardware: All
OS: Linux
low
high
Target Milestone: rc
: ---
Assignee: Virtualization Maintenance
QA Contact: Virtualization Bugs
URL:
Whiteboard:
Keywords:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2010-06-08 10:34 UTC by Keqin Hong
Modified: 2013-01-09 22:42 UTC (History)
7 users (show)

(edit)
Clone Of:
(edit)
Last Closed: 2010-06-08 11:16:27 UTC


Attachments (Terms of Use)

Description Keqin Hong 2010-06-08 10:34:30 UTC
Description of problem:
when using virtio with vhost-net, and when migration is taken under stress of incoming and outgoing packets, migration never ends.

Version-Release number of selected component (if applicable):
qemu-kvm 0.12.1.2-2.71
kernel-2.6.32-33.el6.x86_64

How reproducible:
almost always

CLI:
# /usr/libexec/qemu-kvm -m 2G -smp 2 -drive file=RHEL6.0-64-virtio-0603.1.qcow2,if=none,id=drive-virtio0,boot=on -device virtio-blk-pci,drive=drive-virtio0,id=virtio-blk-pci0,addr=0x3 -netdev tap,id=hostnet0,vhost=on -device virtio-net-pci,netdev=hostnet0,mac=02:00:40:3F:20:20,bus=pci.0,addr=0x4 -boot order=c,menu=on -uuid 11144ecc-d3a1-4d3c-a386-12daf50015f2 -rtc base=utc -rtc-td-hack -no-kvm-pit-reinjection -monitor stdio -cpu qemu64,+sse2 -balloon none -vnc :1

Steps to Reproduce:
1. start guest with vhost=on
2. start netserver on both host and guest
host$ ./netserver -p 5912
guest$ ./netserver -p 5920
3. run netperf from host to guest
host$ while true; do ./netperf -H $guestip -p 5920; done
4. at the same time run netperf from guest to host
guest$ while true; do ./netperf -H $hostip -p 5912; done
5. migrate guest
  
Actual results:
Migration starts but never ends.
1. (qemu) info migrate shows the remaining ram fluctuating around a certain point, as in
"(qemu) info migrate 
Migration status: active
transferred ram: 3187272 kbytes
remaining ram: 39644 kbytes
total ram: 2113920 kbytes
(qemu) info migrate 
Migration status: active
transferred ram: 11353476 kbytes
remaining ram: 42604 kbytes
total ram: 2113920 kbytes
(qemu) info migrate 
Migration status: active
transferred ram: 27299628 kbytes
remaining ram: 39564 kbytes
total ram: 2113920 kbytes
...
"
2. On stop either host side or guest side netperf process, which reduces network stress, migration will soon complete. 

Expected results:
Migration shouldn't proceed in an endless loop; Complete or fail within a certain amount of time.

Comment 2 RHEL Product and Program Management 2010-06-08 10:52:57 UTC
This request was evaluated by Red Hat Product Management for inclusion in a Red
Hat Enterprise Linux major release.  Product Management has requested further
review of this request by Red Hat Engineering, for potential inclusion in a Red
Hat Enterprise Linux Major release.  This request is not yet committed for
inclusion.

Comment 3 Dor Laor 2010-06-08 11:16:27 UTC
It's not a bug, it is the expected behaviour.
In this case libvirt should increase the migration bandwidth or even pause the guest and stop the live part.


Note You need to log in before you can comment on or make changes to this bug.