I have checked the /var/log/vdsm/mom.log and it seems that MOM was actually working:
2019-06-13 07:08:47,690 - mom.GuestMonitor.Thread - INFO - GuestMonitor-node1 starting
2019-06-13 07:09:39,490 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 1048576 to 996147
2019-06-13 07:09:54,658 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 996148 to 946340
2019-06-13 07:10:09,853 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 946340 to 899023
2019-06-13 07:10:25,053 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 899024 to 854072
2019-06-13 07:10:40,233 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 854072 to 811368
2019-06-13 07:10:55,428 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 811368 to 770799
2019-06-13 07:11:10,621 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 770800 to 732260
2019-06-13 07:11:25,827 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 732260 to 695647
2019-06-13 07:11:40,973 - mom.Controllers.Balloon - INFO - Ballooning guest:node1 from 695648 to 660865
2019-06-13 07:12:51,437 - mom.GuestMonitor.Thread - INFO - GuestMonitor-node1 ending
Can someone clarify what exactly does this (from xxxx to yyyy) mean ?
Best Regards,
Strahil Nikolov
В четвъртък, 13 юни 2019 г., 17:27:01 ч. Гринуич+3, Martin Sivak <msivak@redhat.com> написа:
Hi,
iirc the guest agent is not needed anymore as we get almost the same stats from the balloon driver directly.
Ballooning has to be enabled on cluster level though. So that is one thing to check. If that is fine then I guess a more detailed description is needed.
oVirt generally starts ballooning when the memory load gets over 80% of available memory.
The host agent that handles ballooning is called mom and the logs are located in /var/log/vdsm/mom* iirc. It might be a good idea to check whether the virtual machines were declared ready (meaning all data sources we collect provided data).