Bug 1551016

Summary: Creating ports using OVN as a backend doesn't scale
Product: Red Hat OpenStack Reporter: Daniel Alvarez Sanchez <dalvarez>
Component: openvswitchAssignee: Timothy Redaelli <tredaelli>
Status: CLOSED ERRATA QA Contact: Eran Kuris <ekuris>
Severity: high Docs Contact:
Priority: high    
Version: 13.0 (Queens)CC: amuller, apevec, bcafarel, chrisw, dalvarez, ekuris, joflynn, jschluet, lhh, majopela, nyechiel, rhos-maint, srevivo, tfreger, tredaelli
Target Milestone: z2Keywords: Triaged, ZStream
Target Release: 13.0 (Queens)   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: openvswitch-2.9.0-21.el7fdn Doc Type: If docs needed, set a value
Doc Text:
The creation time of OVN ports grew linearly as ports were created. The creation time now remains constant, regardless of the number of ports in the cloud.
Story Points: ---
Clone Of: 1545344 Environment:
Last Closed: 2018-08-29 16:32:35 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On: 1545344, 1622469    
Bug Blocks:    

Comment 1 Daniel Alvarez Sanchez 2018-03-02 14:55:42 UTC
This is greatly improved by this patch [0]. Patch got merged in master and I asked for the backport upstream to branch 2.9 (ie OSP13). Link to the upstream discussion at [1].

[0] https://github.com/openvswitch/ovs/commit/eab138764e3aaa9f98a725a4636dd30e2a41a896
[1] https://mail.openvswitch.org/pipermail/ovs-discuss/2018-February/046149.html

Comment 18 Daniel Alvarez Sanchez 2018-08-09 13:26:13 UTC
If you want to measure the times for creating a port in OVN:

diff --git a/networking_ovn/ml2/mech_driver.py b/networking_ovn/ml2/mech_driver.py
index 2fb7d35..0ebbc0e 100644
--- a/networking_ovn/ml2/mech_driver.py
+++ b/networking_ovn/ml2/mech_driver.py
@@ -27,6 +27,7 @@ from neutron_lib.services.qos import constants as qos_consts
 from oslo_config import cfg
 from oslo_db import exception as os_db_exc
 from oslo_log import log
+from oslo_utils import timeutils
 
 from neutron.common import utils as n_utils
 from neutron.db import provisioning_blocks
@@ -471,7 +472,11 @@ class OVNMechanismDriver(api.MechanismDriver):
         result in the deletion of the resource.
         """
         port = context.current
+        _timer = timeutils.StopWatch()
+        _timer.restart()
         self._ovn_client.create_port(port)
+        _timer.stop()
+        LOG.debug('[*] create_port = %.2f seconds', _timer.elapsed())
         self._notify_dhcp_updated(port['id'])




And then parse the logs:



import sys
import re

PATTERN = re.compile(r'create_port\ =\ (\d+\.\d+)\ seconds')

with open(sys.argv[1]) as logf:
    for line in logf:
        match = PATTERN.findall(line)
        if match:
            print(match[0])

Comment 19 Joanne O'Flynn 2018-08-13 10:32:11 UTC
This bug is marked for inclusion in the errata but does not currently contain draft documentation text. To ensure the timely release of this advisory please provide draft documentation text for this bug as soon as possible.

If you do not think this bug requires errata documentation, set the requires_doc_text flag to "-".


To add draft documentation text:

* Select the documentation type from the "Doc Type" drop down field.

* A template will be provided in the "Doc Text" field based on the "Doc Type" value selected. Enter draft text in the "Doc Text" field.

Comment 20 Timothy Redaelli 2018-08-13 10:49:24 UTC
I only backported the patch, so I think the original creator of the feature should add the doctext

Comment 21 Eran Kuris 2018-08-16 06:07:07 UTC
fix verified on OSP13 2018-08-08.2

openvswitch-2.9.0-54.el7fdp.x86_64

The average time of creation port 0.425

added to the neutron server code:
diff --git a/networking_ovn/ml2/mech_driver.py b/networking_ovn/ml2/mech_driver.py
index 2fb7d35..0ebbc0e 100644
--- a/networking_ovn/ml2/mech_driver.py
+++ b/networking_ovn/ml2/mech_driver.py
@@ -27,6 +27,7 @@ from neutron_lib.services.qos import constants as qos_consts
 from oslo_config import cfg
 from oslo_db import exception as os_db_exc
 from oslo_log import log
+from oslo_utils import timeutils
 
 from neutron.common import utils as n_utils
 from neutron.db import provisioning_blocks
@@ -471,7 +472,11 @@ class OVNMechanismDriver(api.MechanismDriver):
         result in the deletion of the resource.
         """
         port = context.current
+        _timer = timeutils.StopWatch()
+        _timer.restart()
         self._ovn_client.create_port(port)
+        _timer.stop()
+        LOG.debug('[*] create_port = %.2f seconds', _timer.elapsed())
         self._notify_dhcp_updated(port['id'])

collected the logs from all controller node and ran the script:
import sys
import re

PATTERN = re.compile(r'create_port\ =\ (\d+\.\d+)\ seconds')

with open(sys.argv[1]) as logf:
    for line in logf:
        match = PATTERN.findall(line)
        if match:
            print(match[0])

Comment 23 errata-xmlrpc 2018-08-29 16:32:35 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, 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-2018:2573