Actions
action #109190
closedInvalid reusage of VLAN-Tag in multi-machine scenario, when using job posts
Status:
Closed
Priority:
Normal
Assignee:
Category:
Regressions/Crashes
Target version:
Start date:
2022-03-29
Due date:
% Done:
0%
Estimated time:
Description
Observation¶
- The job http://openqa-3.wicked.suse.de/tests/80283#step/before_test/23 show a
eth0: IPv4 duplicate address 10.0.2.11 detected (in use by 52:54:00:12:00:56)!
message - The mac 52:54:00:12:00:56 belongs to http://openqa-3.wicked.suse.de/admin/workers/86 which runs the job http://openqa-3.wicked.suse.de/tests/80207 during that time.
The job 80207 is a multi machine job and the parent is http://openqa-3.wicked.suse.de/tests/80206, which start at 2022-03-29T09:27:13.195681+02:00
and end at 2022-03-29T10:04:05.167885+02:00
, while job 80207 ends at [2022-03-29T10:06:21.469947+02:00]
.
The failing job show the qemu command at: [2022-03-29T10:04:26.913030+02:00] [debug] starting: /usr/bin/qemu-system-x86_64 -vga cirrus -only-migratable...
, thus
it start a qemu instance with the same VLAN which is still used by job 80207.
Simple reproducer, create two parallel boot jobs
id=$(openqa-cli api --host http://openqa-3.wicked.suse.de -X POST jobs 'ARCH=x86_64' 'DISTRI=opensuse' 'FLAVOR=CI' 'MACHINE=x86_64' 'VERSION=Tumbleweed' '_GROUP_ID=0' \
'BOOT_HDD_IMAGE=1' 'DESKTOP=textmode' 'HDD_1=tumbleweed.qcow2' 'KEEP_GRUB_TIMEOUT=1' \
'BACKEND=qemu' 'NICTYPE=tap' 'WORKER_CLASS=tap,qemu_x86_64' \
'SCHEDULE=tests/boot/boot_to_desktop' 'TEST=check_vlan_on_mm_job_parent' | jq -r '.id')
echo "PARENT_ID:$id"
openqa-cli api --host http://openqa-3.wicked.suse.de -X POST jobs 'ARCH=x86_64' 'DISTRI=opensuse' 'FLAVOR=CI' 'MACHINE=x86_64' 'VERSION=Tumbleweed' '_GROUP_ID=0' \
'BOOT_HDD_IMAGE=1' 'DESKTOP=textmode' 'HDD_1=tumbleweed.qcow2' 'KEEP_GRUB_TIMEOUT=1' \
'BACKEND=qemu' 'NICTYPE=tap' 'WORKER_CLASS=tap,qemu_x86_64' \
'SCHEDULE=tests/boot/boot_to_desktop' 'TEST=check_vlan_on_mm_job_child' \
"_PARALLEL_JOBS=$id"
Actions