BTW, I modified ovs-ctl.in manually, skipped inserting the kernel
modules, ovs service start successfully and everything works fine so far.
I am using openvswitch in fedora 17:
# rpm -q openvswitch
openvswitch-1.4.2-1.fc17.x86_64
___
dev mailing list
On 02/06/2013 03:14 AM, Ben Pfaff wrote:
Reported-by: Cong Wang
Signed-off-by: Ben Pfaff
---
Cong, will you verify that this makes the init script work OK for
your kernel with OVS built-in?
I got:
# ovs-dpctl show -voff
ovs-dpctl: processing "off": unknown module "off"
# modprobe openvswit
On Tue, Feb 05, 2013 at 05:07:53PM -0800, Pavithra Ramesh wrote:
> If socket path specified is relative to ovs_rundir(),
> append the directory name to in unix_open.
>
> Signed-off-by: Pavithra Ramesh
Hi Pavithra, I thought that the goal that we discussed in person was
this: if the path is not a
On Tue, Feb 05, 2013 at 09:19:06AM -0800, Ben Pfaff wrote:
> On Tue, Feb 05, 2013 at 10:13:18AM +0900, Simon Horman wrote:
> > On Tue, Feb 05, 2013 at 09:00:35AM +0900, Simon Horman wrote:
> > > On Mon, Feb 04, 2013 at 10:09:20AM -0800, Ben Pfaff wrote:
> > > > On Fri, Jan 25, 2013 at 04:22:07PM +0
Oh, also I think this may introduce a memory leak. I.E. in the case
where connect_path is malloced, it never gets freed.
Ethan
On Tue, Feb 5, 2013 at 5:08 PM, Ethan Jackson wrote:
> This isn't a full review of the patch, but have you considered using
> xasprintf() instead of doing the string man
This isn't a full review of the patch, but have you considered using
xasprintf() instead of doing the string manipulation by hand?
Ethan
On Tue, Feb 5, 2013 at 5:07 PM, Pavithra Ramesh wrote:
> If socket path specified is relative to ovs_rundir(),
> append the directory name to in unix_open.
>
>
If socket path specified is relative to ovs_rundir(),
append the directory name to in unix_open.
Signed-off-by: Pavithra Ramesh
---
lib/stream-unix.c | 12 +++-
1 files changed, 11 insertions(+), 1 deletions(-)
diff --git a/lib/stream-unix.c b/lib/stream-unix.c
index 6ed7648..8764b6c
On Tue, Feb 5, 2013 at 8:53 AM, Rajahalme, Jarno (NSN - FI/Espoo)
wrote
> On Jan 31, 2013, at 3:19 , ext Jesse Gross wrote:
>> On Wed, Jan 30, 2013 at 3:14 AM, Rajahalme, Jarno (NSN - FI/Espoo)
>> wrote:
>>>
>>> Otherwise not so much use of the layer pointers, but
>>> dpif_flow_stats_extract()
>
Thanks for the reviews, I'll merge the last two patches of this series shortly.
Ethan
On Tue, Feb 5, 2013 at 8:55 AM, Ben Pfaff wrote:
> On Mon, Feb 04, 2013 at 07:13:36PM -0800, Ethan Jackson wrote:
>> The autopath action was attempting to achieve functionality similar
>> to the bundle action,
Looks like a massive win to me.
Acked-by: Ethan Jackson
On Fri, Feb 1, 2013 at 10:48 AM, Ben Pfaff wrote:
> It is necessary to whitelist a couple of tests that appear to legitimately
> have WARN messages.
>
> Signed-off-by: Ben Pfaff
> ---
> tests/ofproto-macros.at | 20 +++-
Acked-by: Ethan Jackson
On Fri, Feb 1, 2013 at 10:48 AM, Ben Pfaff wrote:
> It's not meaningful in that case.
>
> Signed-off-by: Ben Pfaff
> ---
> lib/timeval.c |4 ++--
> 1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/lib/timeval.c b/lib/timeval.c
> index d91305c..4ffb7
On Tue, Feb 05, 2013 at 12:56:43PM -0800, Ethan Jackson wrote:
> Since I've dropped the first patch, I think choose_hash_slave() is no longer
> necessary. In this version of the patch, I've removed it.
Looks good, thanks.
___
dev mailing list
dev@openvs
This seems fine to me. The default bond mode has been active-backup
for quite a while, so it may simply make sense to remove this warning
in the post 1.10 release.
Acked-by: Ethan Jackson
On Fri, Feb 1, 2013 at 10:48 AM, Ben Pfaff wrote:
> This avoids a log warning:
>
> bridge|WARN|port bo
Acked-by: Ethan Jackson
On Fri, Feb 1, 2013 at 10:48 AM, Ben Pfaff wrote:
> An interface coming up or going down isn't a big deal.
>
> Signed-off-by: Ben Pfaff
> ---
> lib/bond.c |6 +++---
> 1 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/lib/bond.c b/lib/bond.c
> index
This has annoyed me for a while, thanks.
Acked-by: Ethan Jackson
On Fri, Feb 1, 2013 at 10:48 AM, Ben Pfaff wrote:
> These can happen occasionally in normal circumstances.
>
> Signed-off-by: Ben Pfaff
> ---
> lib/poll-loop.c |4 ++--
> 1 files changed, 2 insertions(+), 2 deletions(-)
>
>
Acked-by: Ethan Jackson
On Fri, Feb 1, 2013 at 10:48 AM, Ben Pfaff wrote:
> Otherwise, if the test bails out before finishing, the test-netflow daemon
> doesn't get killed and the test hangs.
>
> Signed-off-by: Ben Pfaff
> ---
> tests/ofproto-dpif.at |2 +-
> 1 files changed, 1 insertions(
Since I've dropped the first patch, I think choose_hash_slave() is no longer
necessary. In this version of the patch, I've removed it.
---
NEWS |1 +
lib/bond.c | 87 +++-
lib/bond.h |4 +-
of
Upon reflection, I think you're right. I've just dropped this patch.
Ethan
On Tue, Feb 5, 2013 at 8:40 AM, Ben Pfaff wrote:
> On Mon, Feb 04, 2013 at 07:13:34PM -0800, Ethan Jackson wrote:
>> Traditionally, (unless balancing was turned off), the bonding code
>> chose the active slave for all fl
Reported-by: Cong Wang
Signed-off-by: Ben Pfaff
---
Cong, will you verify that this makes the init script work OK for
your kernel with OVS built-in?
Thanks,
Ben.
utilities/ovs-ctl.in |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/utilities/ovs-ctl.in b/utilities
On Tue, Feb 05, 2013 at 10:13:18AM +0900, Simon Horman wrote:
> On Tue, Feb 05, 2013 at 09:00:35AM +0900, Simon Horman wrote:
> > On Mon, Feb 04, 2013 at 10:09:20AM -0800, Ben Pfaff wrote:
> > > On Fri, Jan 25, 2013 at 04:22:07PM +0900, Simon Horman wrote:
> > > > This patch implements use-space da
On Mon, Feb 04, 2013 at 07:13:36PM -0800, Ethan Jackson wrote:
> The autopath action was attempting to achieve functionality similar
> to the bundle action, but was significantly clunkier, more
> difficult to understand, more difficult to use, and less reliable.
> This patch removes it.
>
> Signed
On Jan 31, 2013, at 3:19 , ext Jesse Gross wrote:
> On Wed, Jan 30, 2013 at 3:14 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> Otherwise not so much use of the layer pointers, but
>> dpif_flow_stats_extract()
>> does packet_get_tcp_flags(), which needs the layer pointers.
>> dpif_flow_s
On Mon, Feb 04, 2013 at 07:13:35PM -0800, Ethan Jackson wrote:
> Stable bond mode, along with autopath, were trying to implement
> functionality close to what we get from the bundle action.
> Unfortunately, they are quite clunky, and generally less useful
> than bundle, so they're being removed.
>
On Mon, Feb 04, 2013 at 07:13:34PM -0800, Ethan Jackson wrote:
> Traditionally, (unless balancing was turned off), the bonding code
> chose the active slave for all flows which hadn't previously been
> allocated to another. This worked fine in theory because traffic
> would eventually be balanced
24 matches
Mail list logo
