[ovs-discuss] ovn: OF rules for datapath not getting removed upon ls-del

[Flavio Fernandes]

[cc: Numan and Guru, Ben, Ryan]

Hi folks,

I've been chasing an issue in that OF rules are not properly cleaned up when I remove the datapath that is
associated with a logical switch. The smallest/dumb test I could come up with looks like this:

1) start OVS sandbox  (cd ./utilities && rm -rf ./sandbox && ./ovs-sim)
2) ovn_start, net_add ... create hv1
3) as hv1, create ovs port to be bound to lsp 'foo'
4) ls-add sw0
5) lsp-add port 'foo'
6) undo 3 (ovs port create)
7) undo 5 (lsp create)
8) undo 4 (ls create)
9) look at output of "OF flows" and verify that there is no mentioning of metadata=0x1 anywhere

The exact test steps are here [testSteps].

For the output of a 'good' run, look here [goodRun].
For the output of a 'bad' run, look here [badRun].

Using git bisect, I determined that the issue was very likely introduced here [issueStart].

All in all, I would very much appreciate if:

1) any of you guys "see" the same, and agree this is an issue;
2) know if the changes for "reset flow processing" could be improved so these rules stop 'leaking'.

I suspect there may be other issues, but I figure we start here and work our way up to the
commits in 2.6.

Thank you!

-- flaviof


[testSteps]:   https://gist.github.com/de0bf9e1211267f1afd547aad6072c69 <https://gist.github.com/de0bf9e1211267f1afd547aad6072c69>
[goodRun]: https://gist.github.com/0afeac5b75a0ef6bf015d6868a12a3cd <https://gist.github.com/0afeac5b75a0ef6bf015d6868a12a3cd>
[badRun]: https://gist.github.com/67cfca69abf8ccba1c57b4a5ff08a94e <https://gist.github.com/67cfca69abf8ccba1c57b4a5ff08a94e>


[issueStart]:
f5792c3f36dee70f5c17d03982dce212847024b3 is the first bad commit
commit f5792c3f36dee70f5c17d03982dce212847024b3
Author: Numan Siddique <nusiddiq at redhat.com>
Date:   Thu Aug 11 17:51:39 2016 +0530

    ovn-controller: Reset flow processing after (re)connection to switch

    When ovn-controller reconnects to the ovs-vswitchd, it deletes all the
    OF flows in the switch. It doesn't install the flows again, leaving
    the datapath broken unless ovn-controller is restarted or ovn-northd
    updates the SB DB.

    The reason for this is
      - lflow_reset_processing() is not called after the reconnection
      - the hmap "installed_flows" is not cleared, because of which
        ofctrl_put skips adding the flows to the switch.

    This patch fixes the issue and also adds a test case to test
    this scenario.

    Signed-off-by: Numan Siddique <nusiddiq at redhat.com>
    Signed-off-by: Ben Pfaff <blp at ovn.org>
    Acked-by: Ryan Moats <rmoats at us.ibm.com>

:040000 040000 3687d0bb43c69133c7709f6b22315060354c7253 d5da3e29f654a92acf58fcd215366b5aeb53aef6 M     	ovn
:040000 040000 caf406231eba7541aab62c6bda301f22256e7c76 18f08e67bb4087ea6d05f59356802ef75bc0de81 M     	tests




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20160822/bc868ca4/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20160822/bc868ca4/attachment-0002.sig>