2017-03-10 16:47:59 |
TatyanaGladysheva |
description |
Detailed bug description:
Sometimes UMM_DRC counter doesn't increase after 'virsh destroy/start' of controller node.
Steps to reproduce:
1) Deploy a MOS environment
2) Ensure there are default values of "REBOOT_COUNT" and "COUNTER_RESET_TIME" is configured on a Controller node:
cat /etc/umm.conf
UMM=yes
REBOOT_COUNT=2
COUNTER_RESET_TIME=10
3) Perform reboot of controller node; check value of UMM_DRC variable:
root@node-1:~# sudo reboot
After that make sure that UMM_DRC is equal to 1:
root@node-1:~# cat /var/lib/umm/*
UMM_DRC=1
4) Kill a Libvirt VM which corresponds to Controller node using virsh then start it immediately:
virsh destroy fuel_custom_slave-01
virsh start fuel_custom_slave-01
5) Enter the Controller node and check value of UMM_DRC variable: 'cat /var/lib/umm/*'
Expected results:
UMM_DRC=2 on step 5).
Actual results:
Sometimes UMM_DRC=1 on step 5).
http://paste.openstack.org/show/602246/ - 1 half of file - issue is not reproduced; 2 half of file - issue is reproduced.
Reproducibility:
50%
Impact:
For example, to get umm mode we need to make additional 'virsh destroy/start' of node, although this count is defined in REBOOT_COUNT value in /etc/umm.conf.
Description of the environment:
MOS 9.2
fuel-library9.0-9.0.0-1.mos8740.noarch
fuel-library9.0-9.0.0-1.mos8727.noarch |
Detailed bug description:
Sometimes UMM_DRC counter doesn't increase its value after 'virsh destroy/start' of controller node.
Steps to reproduce:
1) Deploy a MOS environment
2) Ensure there are default values of "REBOOT_COUNT" and "COUNTER_RESET_TIME" is configured on a Controller node:
cat /etc/umm.conf
UMM=yes
REBOOT_COUNT=2
COUNTER_RESET_TIME=10
3) Perform reboot of controller node; check value of UMM_DRC variable:
root@node-1:~# sudo reboot
After that make sure that UMM_DRC is equal to 1:
root@node-1:~# cat /var/lib/umm/*
UMM_DRC=1
4) Kill a Libvirt VM which corresponds to Controller node using virsh then start it immediately:
virsh destroy fuel_custom_slave-01
virsh start fuel_custom_slave-01
5) Enter the Controller node and check value of UMM_DRC variable: 'cat /var/lib/umm/*'
Expected results:
UMM_DRC=2 on step 5).
Actual results:
Sometimes UMM_DRC=1 on step 5).
http://paste.openstack.org/show/602246/ - 1 half of file - issue is not reproduced; 2 half of file - issue is reproduced.
Reproducibility:
50%
Impact:
For example, to get umm mode we need to make additional 'virsh destroy/start' of node, although this count is defined in REBOOT_COUNT value in /etc/umm.conf.
Description of the environment:
MOS 9.2
fuel-library9.0-9.0.0-1.mos8740.noarch
fuel-library9.0-9.0.0-1.mos8727.noarch
Additional information:
http://paste.openstack.org/show/602197/
To get umm mode (9.2 + mu1, fuel-library9.0-9.0.0-1.mos8740.noarch):
If we start make destroy/start's when UMM_DRC counter is cleaned ("/var/lib/umm/*: No such file or directory" case) then we need at least 3 subsequent destroy-start's to get umm mode, but REBOOT_COUNT=2 in /etc/umm.conf. So we need 1 extra destroy/start to get umm mode.
Therefore to fix it we should set UMM_DRC=1 after COUNTER_RESET_TIME expires instead of cleaning of UMM_DRC. |
|