Bug 2011178 - networking-ovn high RSS memory usage
Summary: networking-ovn high RSS memory usage
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Red Hat OpenStack
Classification: Red Hat
Component: python-networking-ovn
Version: 16.2 (Train)
Hardware: Unspecified
OS: Unspecified
high
high
Target Milestone: z2
: 16.2 (Train on RHEL 8.4)
Assignee: Lucas Alvares Gomes
QA Contact: Eran Kuris
URL:
Whiteboard:
Depends On:
Blocks: 2126059
TreeView+ depends on / blocked
 
Reported: 2021-10-06 08:22 UTC by anil venkata
Modified: 2022-09-12 10:10 UTC (History)
7 users (show)

Fixed In Version: python-networking-ovn-7.4.2-2.20220113214852.a2eba10.el8ost
Doc Type: No Doc Update
Doc Text:
Clone Of:
: 2126059 (view as bug list)
Environment:
Last Closed: 2022-03-23 22:11:48 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
OpenStack gerrit 812805 0 None MERGED [ovn] Stop monitoring the SB MAC_Binding table to reduce mem footprint 2021-10-20 13:20:38 UTC
Red Hat Issue Tracker OSP-10217 0 None None None 2021-11-15 12:39:13 UTC
Red Hat Product Errata RHBA-2022:1001 0 None None None 2022-03-23 22:12:12 UTC

Comment 2 anil venkata 2021-10-10 08:09:18 UTC
We have a great improvement (i.e RSS memory reduced from 73.5Gib to 8.35Gib) with @Daniel Alvarez Sanchez  mac binding patch [1]
When we have 1.18 million MAC binding entries for 4k VMs, 2.4k floating ips and 800 routers, without the patch neutron-server is consuming 73.5Gib.  With the Daniel's patch [1] neutron-server is taking 8.35Gib.
No performance impact with the patch i.e both the tests have taken 8 hours 30 minutes to complete.

Grafana links for the testing without the fix [1]
RSS memory

https://snapshot.raintank.io/dashboard/snapshot/z7znO1rASYyJzbocjPXFTI9x4V5eTF1F

OSP DB resources during the testing

https://snapshot.raintank.io/dashboard/snapshot/rCmGlIjKahDGDS8S2DPRUElsQ4zqu2KZ

OVN DB resources without MAC_Binding and logical flows (note: collectd is calling ovsdb-client for these metrics)

https://snapshot.raintank.io/dashboard/snapshot/mcy703BkDVALyaLGzYqf1jkJ56U8rpLR

MAC_Binding and logical_flows count (note: collectd is calling ovsdb-client for these metrics)

https://snapshot.raintank.io/dashboard/snapshot/SewHBjAZDoW0tuG4N7N79nUw8xMWAN1D




Grafana links for the testing with the mac binding fix [1]
MAC binding fix RSS memory

https://snapshot.raintank.io/dashboard/snapshot/pbXcD0kABoJ8L4BX7MU8x41RJfJMORtj

OSP DB resources during the testing

https://snapshot.raintank.io/dashboard/snapshot/VKZ75F20L9uuO8YSXD6XkKMzuwXj8pz4

OVN DB resources without MAC_Binding and logical flows (note: collectd is calling ovsdb-client for these metrics)

https://snapshot.raintank.io/dashboard/snapshot/a9V1zcDxiM0B62bq86CsFcj0SE323WiQ

MAC_Binding and logical_flows count (note: collectd is calling ovsdb-client for these metrics)

https://snapshot.raintank.io/dashboard/snapshot/4I46csupL1WCqVINLM66nTVeeUgt9sBg

Comment 9 anil venkata 2022-02-22 09:24:32 UTC
We used the puddle RHOS-16.2-RHEL-8-20220201.n.1 in our scale testing and we could see neutron server using 8G RSS memory for 4K VMs.
SO the fix is working as expected.

Comment 10 Eran Kuris 2022-02-22 09:27:22 UTC
Following comment 9 I am setting this bz to verified 
 thanks a lot, Anil for helping with that issue.

Comment 15 errata-xmlrpc 2022-03-23 22:11:48 UTC
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Release of components for Red Hat OpenStack Platform 16.2.2), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2022:1001


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