Hi team,
I have tested a recent integration/distribution build off the stable/beryllium
branch and can confirm that patch 40744 fixes bug 5582, save for the following
two caveats which I would appreciate your feedback on:
1) Flow durations are now “reset” when the switch is reconnected after
doing a del-manager – is this ok?
a. E.g.
root@mininet-vm:/root/openvswitch-2.5.0# ovs-vsctl show
708f2ebf-3168-4e0c-8138-cdccb7f1c4e8
Manager "tcp:192.168.56.101:6640"
is_connected: true
Bridge br-ex
Controller "tcp:192.168.56.101:6653"
is_connected: true
fail_mode: secure
Port br-ex
Interface br-ex
type: internal
Bridge br-int
Controller "tcp:192.168.56.101:6653"
is_connected: true
fail_mode: secure
Port br-int
Interface br-int
type: internal
ovs_version: "2.5.0"
root@mininet-vm:/root/openvswitch-2.5.0# ovs-vsctl del-manager
root@mininet-vm:/root/openvswitch-2.5.0# ovs-ofctl dump-flows -O Openflow13
br-ex
OFPST_FLOW reply (OF1.3) (xid=0x2):
cookie=0x0, duration=1743.438s, table=0, n_packets=0, n_bytes=0, dl_type=0x88cc
actions=CONTROLLER:65535
cookie=0x0, duration=1743.438s, table=0, n_packets=0, n_bytes=0, priority=0
actions=NORMAL
root@mininet-vm:/root/openvswitch-2.5.0# ovs-vsctl set-manager
"tcp:192.168.56.101:6640"
root@mininet-vm:/root/openvswitch-2.5.0# ovs-ofctl dump-flows -O Openflow13
br-ex
OFPST_FLOW reply (OF1.3) (xid=0x2):
cookie=0x0, duration=14.952s, table=0, n_packets=0, n_bytes=0, dl_type=0x88cc
actions=CONTROLLER:65535
cookie=0x0, duration=14.939s, table=0, n_packets=0, n_bytes=0, priority=0
actions=NORMAL
2) br-int flow reconciliation behaviour has changed – br-int flows are no
longer pushed down to the switch upon reconnection (see email
https://lists.opendaylight.org/pipermail/ovsdb-dev/2016-June/003389.html)
Thanks,
Bertrand
From: [email protected]
[mailto:[email protected]] On Behalf Of
Bertrand Low
Sent: 28 June 2016 16:36
To: Shuva Jyoti Kar <[email protected]>; Muthukumaran K
<[email protected]>; Anil Vishnoi <[email protected]>
Cc: [email protected]
Subject: Re: [openflowplugin-dev] current status of Openflow Plugin
Reconciliation
Hi Shuva, thanks, appreciate the reply.
I understand that bug 6085 is regarding modification of the same flow, and I
agree that the patch works to add the desired flow onto the switch. What I’m
concerned with is if there are use cases in which FLOW_MOD_ADD with
overlapping-match=false is not equivalent to MODIFY/MODIFY_STRICT. I’m not
aware of all the use cases where MD-SAL flows are used to craft FLOW_MOD
commands to the switch, but wanted to make sure that all use cases can still be
satisfied now that both UpdatedFlow and AddFlowInput essentially send a
FLOW_MOD_ADD to the switch. i.e. no use case should rely on sending a
MODIFY(_STRICT) to the switch. Would appreciate your insight into this.
I will confirm as soon as I can regarding the 40744 patch fix for bug 5582.
Thanks,
Bertrand
From: Shuva Jyoti Kar [mailto:[email protected]]
Sent: 27 June 2016 22:52
To: Bertrand Low <[email protected]<mailto:[email protected]>>;
Muthukumaran K
<[email protected]<mailto:[email protected]>>; Anil Vishnoi
<[email protected]<mailto:[email protected]>>
Cc:
[email protected]<mailto:[email protected]>
Subject: RE: [openflowplugin-dev] current status of Openflow Plugin
Reconciliation
Hi Betrand,
Please find my answers inline and do let me know in case you have any thoughts.
Thanks
Shuva
From: Bertrand Low [mailto:[email protected]]
Sent: Tuesday, June 28, 2016 6:05 AM
To: Shuva Jyoti Kar; Muthukumaran K
Cc:
[email protected]<mailto:[email protected]>
Subject: RE: [openflowplugin-dev] current status of Openflow Plugin
Reconciliation
Hi Shuva,
Yes, my patch provides logic to determine if a FLOW_MOD_ADD should be sent
instead of a MODIFY_STRICT.
For the patch you mentioned at https://git.opendaylight.org/gerrit/#/c/40744/,
I’m quite confident that this will address the bug 5582 issue as well. However,
I’m blocked on confirming it due to the following issue:
https://git.opendaylight.org/gerrit/#/c/40871/
[sk] okii. We can wait for the fix to be in.
While I believe that https://git.opendaylight.org/gerrit/#/c/40744/ will also
resolve bug 5582, I’m concerned about the implications of that patch. Can you
explain the reasoning behind it? Thanks. It would appear to me that the change
to FlowConvertor essentially removes any logical flow update. i.e. now all
UpdatedFlows will be sent as FLOW_MOD_ADD commands. What happens if the
controller wishes to truly perform a modification of an existing flow (which
can leave existing cookie, idle_timeout, hard_timeout, importance, flags,
counters, and duration fields unchanged), and not an add? My patch tried to
address this situation by only sending the FLOW_MOD_ADD if the flow did not
already exist on the ovs-switch. Otherwise, a MODIFY would still be sent.
[sk] the bug 6085 aims to address the fact that if we install a flow and
switch doesn't accept the flow (because of an erroneous parameter) and then
user corrects the flow by modifying the “same” flow in the CDS, the flow won't
get installed on the switch.
FLOW_MOD_ADD with overlapping-match(CHECK_OVERLAP) flag set to false, copies
the flow stats (e.g packet count) for the new flow, so I donot think that is a
concern. So that should take care.
Regarding the bug (https://bugs.opendaylight.org/show_bug.cgi?id=6085) that
https://git.opendaylight.org/gerrit/#/c/40744/ was addressing, I wonder if
perhaps we can have some value validation at the time of the user installing
the flow into the config datastore instead, and disallow the addition if the
desired flow’s values do not pass validation.
Thanks,
Bertrand
From: Shuva Jyoti Kar [mailto:[email protected]]
Sent: 24 June 2016 06:56
To: Bertrand Low <[email protected]<mailto:[email protected]>>
Cc:
[email protected]<mailto:[email protected]>
Subject: RE: [openflowplugin-dev] current status of Openflow Plugin
Reconciliation
Hi Betrand,
As I understand , the patch addresses the fact that we require a FLOW_MOD ADD
instead of MODIFY_STRICT . Am I correct ?
If so could you please check this new patch which has gone in -
https://git.opendaylight.org/gerrit/#/c/40744/ on the master and
https://git.opendaylight.org/gerrit/#/c/40510/ on stable/Be
and see if you are still observing the problem.
-Br, shuva
From: Bertrand Low
Sent: Friday, June 24, 2016 1:40 AM
To: Shuva Jyoti Kar
Cc:
[email protected]<mailto:[email protected]>
Subject: Re: [openflowplugin-dev] current status of Openflow Plugin
Reconciliation
Thanks Shuva,
perhaps we can add this on the agenda for discussion at the next ofplugin call.
In the mean time, I have submitted a new patch for bug 5582
(https://git.opendaylight.org/gerrit/#/c/38895/4) - would appreciate your
comments.
Are any of the remaining reconciliation scenarios tracked on bugzilla?
thanks,
Bertrand
________________________________
From: Shuva Jyoti Kar
<[email protected]<mailto:[email protected]>>
Sent: Wednesday, June 22, 2016 6:41:45 PM
To: Bertrand Low
Cc:
[email protected]<mailto:[email protected]>
Subject: Re: [openflowplugin-dev] current status of Openflow Plugin
Reconciliation
Yes Bertrand. Lets discuss more over the ofplugin call today.
On Thu, Jun 23, 2016 at 6:20 AM, Bertrand Low
<[email protected]<mailto:[email protected]>> wrote:
Hi there,
I'm currently working on bug 5582 which deals with flow reconciliation. Does
anyone have the current status of where Reconciliation is for OpenFlow plugin?
i.e. Are there scenarios that still need to be addressed? I'm interested to
contributing more to this area. In the mean time, I'll check out bugzilla for
related bugs.
I found the following wikis, but they look like they can be updated:
https://wiki.opendaylight.org/view/OpenDaylight_OpenFlow_Plugin:Reconciliation
https://wiki.opendaylight.org/view/OpenDaylight_OpenFlow_Plugin:Backlog:Node_Status_Reconciliation
Thanks,
Bertrand
::DISCLAIMER:: The contents of this e-mail and any attachment(s) are
confidential and intended for the named recipient(s) only. E-mail transmission
is not guaranteed to be secure or error-free as information could be
intercepted, corrupted, lost, destroyed, arrive late or incomplete, or may
contain viruses in transmission. The e mail and its contents (with or without
referred errors) shall therefore not attach any liability on the originator or
HCL or its affiliates. Views or opinions, if any, presented in this email are
solely those of the author and may not necessarily reflect the views or
opinions of HCL or its affiliates. Any form of reproduction, dissemination,
copying, disclosure, modification, distribution and / or publication of this
message without the prior written consent of authorized representative of HCL
is strictly prohibited. If you have received this email in error please delete
it and notify the sender immediately. Before opening any email and/or
attachments, please check them for viruses and other defects..
_______________________________________________
openflowplugin-dev mailing list
[email protected]
https://lists.opendaylight.org/mailman/listinfo/openflowplugin-dev
