Try using perf to look at where system is spending it's time.
On Wed, Apr 10, 2013 at 3:55 PM, SDN learn wrote:
> Hi,
>
> I'm running OVS-1.4 release on a low frequency x86 machine. I see that
> performance doesn't seem to scale linearly from 1 to 2 to 4 cores.
>
> I have three PC setup: PC1 is g
Thanks, applied to master and branch-1.10.
On Wed, Apr 10, 2013 at 05:43:13PM -0700, Ansis Atteka wrote:
> For all these 5 patches:
>
> Acked-by: Ansis Atteka
>
>
> On Wed, Apr 10, 2013 at 4:25 PM, Ben Pfaff wrote:
>
> > v1->v2: Don't exit with a fatal error if one "add-remote"s an invalid
>
For all these 5 patches:
Acked-by: Ansis Atteka
On Wed, Apr 10, 2013 at 4:25 PM, Ben Pfaff wrote:
> v1->v2: Don't exit with a fatal error if one "add-remote"s an invalid
> remote.
>
> Ben Pfaff (5):
> sset: New function sset_sort().
> ovsdb-server: Refactor parsing of remote names to avoi
On Wed, Apr 10, 2013 at 09:21:23AM -0700, Jesse Gross wrote:
> On Wed, Apr 10, 2013 at 2:16 AM, Simon Horman wrote:
> > On Tue, Apr 09, 2013 at 08:44:02AM -0700, Jesse Gross wrote:
> >> On Tue, Apr 9, 2013 at 12:50 AM, Simon Horman wrote:
> >> > On Mon, Apr 08, 2013 at 06:46:29PM -0700, Jesse Gro
An earlier commit changed the Open vSwitch startup scripts so that they
connect to remote managers only after ovs-vswitchd does its initial
configuration, as signaled by ovs-vswitchd detaching from its parent
process. However, a race window remains, because ovs-vswitchd detaching
does not mean tha
This will make it possible, in later commits, to make ovsdb-server connect
to OVS managers only after ovs-vswitchd has completed its initial
configuration.
Signed-off-by: Ben Pfaff
---
ovsdb/ovsdb-server.1.in | 28 ++
ovsdb/ovsdb-server.c| 94 +
Until now, ovs-ctl has started ovsdb-server with the full set of remote
managers configured. This means that ovsdb-server immediately connects to
these managers, before ovs-vswitchd even starts. Because the Open vSwitch
schema has several ephemeral columns, there will be considerable startup
chur
v1->v2: Don't exit with a fatal error if one "add-remote"s an invalid
remote.
Ben Pfaff (5):
sset: New function sset_sort().
ovsdb-server: Refactor parsing of remote names to avoid ovs_fatal().
ovsdb-server: Add commands for adding and removing remotes at
runtime.
ovs-ctl: Connect to r
This will have its first caller in an upcoming commit.
Signed-off-by: Ben Pfaff
---
lib/sset.c | 36 +++-
lib/sset.h |4 +++-
2 files changed, 38 insertions(+), 2 deletions(-)
diff --git a/lib/sset.c b/lib/sset.c
index c5b616f..443538d 100644
--- a/lib/sset
The current users of parse_db_column() are content to terminate with a
fatal error if parsing fails. An upcoming commit requires more flexibility,
so this commit refactors parse_db_column() to make this possible.
Signed-off-by: Ben Pfaff
---
ovsdb/ovsdb-server.c | 97 +
On Wed, Apr 10, 2013 at 01:42:29PM -0700, Ansis Atteka wrote:
> One thing I noticed - if you will try to add invalid remote from "db:",
> then parse_db_column() might fail and call ovs_fatal(). Seems a little
> harsh...
Oops.
Here's an incremental for the commit that adds add-remote to fix
that.
Currently, when we upgrade openvswitch packages, we do restart
of userspace daemons. This does not replace the kernel module.
But almost everytime, we want to also use the new kernel module
that comes with the new version. This means that we need to
manually do a "force-reload-kmod". This step, al
If we install the userspace packages first, without a kernel module
present in the correct path, we get an error because package install
automatically tries to start the userspace daemons and we look for
a kernel module to load.
Installing a kernel module package first and then following it by
use
Hi,
I'm running OVS-1.4 release on a low frequency x86 machine. I see that
performance doesn't seem to scale linearly from 1 to 2 to 4 cores.
I have three PC setup: PC1 is generating netperf traffic, PC2 runs OVS, PC3
is terminating traffic. OVS is bridging between two ethernet ports.
The datapa
On Wed, Apr 10, 2013 at 3:19 PM, Ben Pfaff wrote:
> On Wed, Apr 10, 2013 at 12:25:14PM -0700, Ansis Atteka wrote:
> > On Wed, Apr 10, 2013 at 10:35 AM, Ben Pfaff wrote:
> >
> > > This will make it possible, in later commits, to make ovsdb-server
> connect
> > > to OVS managers only after ovs-vsw
On Wed, Apr 10, 2013 at 12:25:14PM -0700, Ansis Atteka wrote:
> On Wed, Apr 10, 2013 at 10:35 AM, Ben Pfaff wrote:
>
> > This will make it possible, in later commits, to make ovsdb-server connect
> > to OVS managers only after ovs-vswitchd has completed its initial
> > configuration.
> >
> > Sign
On Wed, Apr 10, 2013 at 1:34 PM, Andy Zhou wrote:
>
> Signed-off-by: Andy Zhou
> ---
> datapath/datapath.h |1 -
> 1 file changed, 1 deletion(-)
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Looks good to me.
One thing I noticed - if you will try to add invalid remote from "db:",
then parse_db_column() might fail and call ovs_fatal(). Seems a little
harsh...
On Wed, Apr 10, 2013 at 10:35 AM, Ben Pfaff wrote:
> An earlier commit changed the Open vSwitch startup scripts so that they
Signed-off-by: Andy Zhou
---
datapath/datapath.h |1 -
1 file changed, 1 deletion(-)
diff --git a/datapath/datapath.h b/datapath/datapath.h
index 9bc98fb..ee40d8f 100644
--- a/datapath/datapath.h
+++ b/datapath/datapath.h
@@ -61,7 +61,6 @@ struct dp_stats_percpu {
* struct datapath - data
On Wed, Apr 10, 2013 at 10:35 AM, Ben Pfaff wrote:
> Until now, ovs-ctl has started ovsdb-server with the full set of remote
> managers configured. This means that ovsdb-server immediately connects to
> these managers, before ovs-vswitchd even starts. Because the Open vSwitch
> schema has sever
On Wed, Apr 10, 2013 at 01:09:22PM -0700, Ethan Jackson wrote:
> The list of epoll events contains references to channels which may
> be stale when one of those channels is deleted. The safest thing
> to do is simply refresh epoll() whenever a channel is deleted.
>
> Bug #16057.
> Signed-off-by:
The list of epoll events contains references to channels which may
be stale when one of those channels is deleted. The safest thing
to do is simply refresh epoll() whenever a channel is deleted.
Bug #16057.
Signed-off-by: Ethan Jackson
---
lib/dpif-linux.c |1 +
1 file changed, 1 insertion(
On Wed, Apr 10, 2013 at 12:00:52PM -0700, Gurucharan Shetty wrote:
> Signed-off-by: Gurucharan Shetty
Looks good to me. (I wonder how this odd function got in here?)
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Wed, Apr 10, 2013 at 10:35 AM, Ben Pfaff wrote:
> This will have its first caller in an upcoming commit.
>
> Signed-off-by: Ben Pfaff
> ---
> lib/sset.c | 36 +++-
> lib/sset.h |4 +++-
> 2 files changed, 38 insertions(+), 2 deletions(-)
>
Looks good to
On Wed, Apr 10, 2013 at 10:35 AM, Ben Pfaff wrote:
> This will make it possible, in later commits, to make ovsdb-server connect
> to OVS managers only after ovs-vswitchd has completed its initial
> configuration.
>
> Signed-off-by: Ben Pfaff
> ---
> ovsdb/ovsdb-server.1.in | 28 ++
Signed-off-by: Gurucharan Shetty
---
utilities/ovs-ctl.in |6 +-
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/utilities/ovs-ctl.in b/utilities/ovs-ctl.in
index 5ad5f26..4abcad8 100755
--- a/utilities/ovs-ctl.in
+++ b/utilities/ovs-ctl.in
@@ -30,7 +30,7 @@ done
## start ##
The following two commands results in a ovs-vsctl segfault.
ovs-vsctl -vfatal_signal:off --timeout=0 wait-until \
Open_vswitch . external_ids:blah="1"
/etc/init.d/openvswitch-switch restart
This patch fixes the segfault by properly setting the global
varibale, the_idl_txn to NULL when the underlyi
On Wed, Apr 10, 2013 at 11:40 AM, Pravin Shelar wrote:
> On Wed, Apr 10, 2013 at 9:01 AM, Jesse Gross wrote:
>> By the way, I think we may be able to make the workqueue compatibility
>> code conditional on version again. The comment says that we were
>> having problems with waiting on genl mutex
On Wed, Apr 10, 2013 at 10:32 AM, Jesse Gross wrote:
> On Wed, Apr 10, 2013 at 9:01 AM, Jesse Gross wrote:
> > On Tue, Apr 9, 2013 at 9:04 PM, Pravin Shelar
> wrote:
> >> On Tue, Apr 9, 2013 at 6:41 PM, Jesse Gross wrote:
> >>> I think there is potentially another race condition here: if a
> >
On Wed, Apr 10, 2013 at 9:01 AM, Jesse Gross wrote:
> On Tue, Apr 9, 2013 at 9:04 PM, Pravin Shelar wrote:
> > On Tue, Apr 9, 2013 at 6:41 PM, Jesse Gross wrote:
> >> On Tue, Apr 9, 2013 at 3:03 PM, Pravin B Shelar
> >> wrote:
> >> > diff --git a/datapath/datapath.h b/datapath/datapath.h
> >>
An earlier commit changed the Open vSwitch startup scripts so that they
connect to remote managers only after ovs-vswitchd does its initial
configuration, as signaled by ovs-vswitchd detaching from its parent
process. However, a race window remains, because ovs-vswitchd detaching
does not mean tha
Until now, ovs-ctl has started ovsdb-server with the full set of remote
managers configured. This means that ovsdb-server immediately connects to
these managers, before ovs-vswitchd even starts. Because the Open vSwitch
schema has several ephemeral columns, there will be considerable startup
chur
This will have its first caller in an upcoming commit.
Signed-off-by: Ben Pfaff
---
lib/sset.c | 36 +++-
lib/sset.h |4 +++-
2 files changed, 38 insertions(+), 2 deletions(-)
diff --git a/lib/sset.c b/lib/sset.c
index c5b616f..443538d 100644
--- a/lib/sset
This will make it possible, in later commits, to make ovsdb-server connect
to OVS managers only after ovs-vswitchd has completed its initial
configuration.
Signed-off-by: Ben Pfaff
---
ovsdb/ovsdb-server.1.in | 28 +++
ovsdb/ovsdb-server.c| 68
On Wed, Apr 10, 2013 at 9:01 AM, Jesse Gross wrote:
> On Tue, Apr 9, 2013 at 9:04 PM, Pravin Shelar wrote:
>> On Tue, Apr 9, 2013 at 6:41 PM, Jesse Gross wrote:
>>> I think there is potentially another race condition here: if a
>>> notification comes after we retrieve the value of count but befo
On Wed, Apr 10, 2013 at 2:16 AM, Simon Horman wrote:
> On Tue, Apr 09, 2013 at 08:44:02AM -0700, Jesse Gross wrote:
>> On Tue, Apr 9, 2013 at 12:50 AM, Simon Horman wrote:
>> > On Mon, Apr 08, 2013 at 06:46:29PM -0700, Jesse Gross wrote:
>> >> On Sun, Apr 7, 2013 at 11:43 PM, Simon Horman wrote:
On Wed, Apr 10, 2013 at 3:02 AM, SDN learn wrote:
> Hi Jesse,
>
> I checked the latest trunk code, but couldn't locate single lookup related
> changes...maybe I missed something. Is the code related to single lookup
> table available now?
Yes, it's in master and branch-1.10. Look for code using
On Tue, Apr 9, 2013 at 9:04 PM, Pravin Shelar wrote:
> On Tue, Apr 9, 2013 at 6:41 PM, Jesse Gross wrote:
>> On Tue, Apr 9, 2013 at 3:03 PM, Pravin B Shelar
>> wrote:
>> > diff --git a/datapath/datapath.h b/datapath/datapath.h
>> > index 9bc98fb..58d7169 100644
>> > --- a/datapath/datapath.h
>>
Hi Jesse,
I checked the latest trunk code, but couldn't locate single lookup related
changes...maybe I missed something. Is the code related to single lookup
table available now?
Thanks
Marty!
On Mon, Feb 11, 2013 at 8:21 AM, Jesse Gross wrote:
> On Sun, Feb 10, 2013 at 3:02 AM, SDN learn w
On Tue, Apr 09, 2013 at 08:44:02AM -0700, Jesse Gross wrote:
> On Tue, Apr 9, 2013 at 12:50 AM, Simon Horman wrote:
> > On Mon, Apr 08, 2013 at 06:46:29PM -0700, Jesse Gross wrote:
> >> On Sun, Apr 7, 2013 at 11:43 PM, Simon Horman wrote:
> >> > diff --git a/datapath/actions.c b/datapath/actions.
40 matches
Mail list logo
