ovs-dpdk runs out of mem on StarlingX Simplex baremetal

[Cindy Xie]

Brief Description
-----------------
After StarlingX installed in bare metal server with Simplex config, mem is quickly run out with OVS process OVS-vswitch service. This bug was reported by 99Cloud.

Severity: Critical

System Configuration
It's installed as Simplex on bare metal system. With HW config:
HW config:
- CPU: 56c
- Memory: 188GB
- Hard Disck: 2400G SAS RAID1*1/480G SSD RAID1*1
- OS: bootimage0727.iso

[Ghada Khalil (gkhalil)]

Note that this issue was not seen in testing done by Wind River. Typically, we use Xeon-based hardware (Haswell, Broadwell, Skylake) without any issue.

[LiKai (li-kai-i)]

Our simplex environment had been re-isntalled with multi-nodes node. So this issue can not be reproduced in the near future. We will report it if we met it again.

[Kailun Qin (kailun.qin)]

Waiting for LiKai's feedbacks on whether it can be reproduced with multi-node (Robson) deployment.

[Ghada Khalil (gkhalil)]

LiKai, Is there any further update on this issue? If you're not able to reproduce the issue at this time, we will mark this bug as Expired as we are nearing our release freeze date. You can open a new bug in the future if you encounter the issue again.

[Kailun Qin (kailun.qin)]

Based on the meeting with CUC/99Cloud (reporter) on 9/21,
1) they no longer keep the baremetal simplex environment where this issue is initially reported, so no further reproduction or debug is possible;
2) they are currently adopting a multi-node VE (StarlingX Cloud with Controller Storage Virtual Environment) on a physical server for testing;
3) this issue cannot be reproduced w/ their current deployment (multi-node VE);
4) they don't think this will block their subsequent testing and they suppose it is OK to let go the issue for now.

Let's mark this bug as Expired. Please kindly open a new bug in the future if you encounter the issue again.