[ovs-dev] [PATCH v3] Dynamically reconnect ovn-controller if ovn-remote value changes

Ryan Moats rmoats at us.ibm.com
Tue Apr 5 15:20:52 UTC 2016


[snipping for bandwidth]

Lance Richardson <lrichard at redhat.com> wrote on 04/05/2016 09:36:06 AM:

> > diff --git a/lib/ovsdb-idl.c b/lib/ovsdb-idl.c
> > index 3ab05a3..24c536d 100644
> > --- a/lib/ovsdb-idl.c
> > +++ b/lib/ovsdb-idl.c
> > @@ -272,6 +272,22 @@ ovsdb_idl_create(const char *remote, const struct
> > ovsdb_idl_class *class,
> >      return idl;
> >  }
> >
> > +/* chagnes the remote and creates a new session. */
>
> s/chagnes/Changes/

Oops... Done...

> > diff --git a/tests/ovn-controller.at b/tests/ovn-controller.at
> > index 4bee3ca..aa04d4a 100644
> > --- a/tests/ovn-controller.at
> > +++ b/tests/ovn-controller.at
> > @@ -120,6 +120,21 @@ check_patches \
> >      'br-int patch-quux-to-baz patch-baz-to-quux' \
> >      'br-int patch-baz-to-quux patch-quux-to-baz'
> >
> > +# Create an empty database, serve it and switch to it
> > +# and verify that the OVS patch ports disappear
> > +# then put it back and verify that they reappear
> > +
> > +#on_exit 'kill `cat $ovs_base/ovn-sb/ovsdb-server-2.pid`'
>
> This line is commented out, should it be deleted? BTW, using
> OVS_APP_EXIT_AND_WAIT and friends is the preferred way to clean
> up since killing the process will cause code coverage information
> to be lost (and worse, coverage data can be corrupted if the process
> is killed while in the midst of writing coverage data to disk).

Yeah, that was a typo - it needs to be uncommented.
This is in place in case something goes wrong with the test, and
further, I don't see any use of OVS_APP_EXIT_AND_WAIT when setting
up on_exit in any of the current tests.

Updated patch set coming...

Ryan



More information about the dev mailing list