Joe, based on Ethan's feedback I'll drop this for now unless you prefer
otherwise.
On Tue, Feb 25, 2014 at 05:37:05PM -0800, Ethan Jackson wrote:
> I'm not sure this patch makes sense in the current code. If a
> revalidation failed, that's probably because the rules used to create
> the flow chan
I'm not sure this patch makes sense in the current code. If a
revalidation failed, that's probably because the rules used to create
the flow changed, in which case we'd be accounting packets to the
wrong rules. This is arguable worse than missing the packets all
together.
Once we change the code
A flow's statistics could be partially lost in the following situation:
* A flow is dumped from the datapath
* Some traffic hits that flow
* Revalidation fails for that flow
Previously, we would delete such a flow without fetching the latest
statistics for it, causing the intermediate statistics t
Woops, I had the continue in my git index, so the testsuite passed--but I
forgot to squash it onto this patch. You're right, we definitely need the
continue.
On 25 February 2014 17:06, Ben Pfaff wrote:
> On Tue, Feb 25, 2014 at 05:02:02PM -0800, Joe Stringer wrote:
> > A flow's statistics could
On Tue, Feb 25, 2014 at 05:02:02PM -0800, Joe Stringer wrote:
> A flow's statistics could be partially lost in the following situation:
> * A flow is dumped from the datapath
> * Some traffic hits that flow
> * Revalidation fails for that flow
>
> Previously, we would delete such a flow without fe
A flow's statistics could be partially lost in the following situation:
* A flow is dumped from the datapath
* Some traffic hits that flow
* Revalidation fails for that flow
Previously, we would delete such a flow without fetching the latest
statistics for it, causing the intermediate statistics t
Yep, will do :-)
Jarno
> On Feb 24, 2014, at 10:44 AM, Ben Pfaff wrote:
>
> You want to push it for Joe?
>
>> On Mon, Feb 24, 2014 at 10:37:04AM -0800, Jarno Rajahalme wrote:
>> Acked-by: Jarno Rajahalme
>>
>>> On Feb 24, 2014, at 10:20 AM, Joe Stringer wrote:
>>>
>>> Commit 8bfd0fda mis
Thanks, I applied both of these to master.
(I avoided making these mistakes in the backport, so no change is
necessary on branch-2.1.)
On Tue, Feb 25, 2014 at 04:39:52PM -0800, Joe Stringer wrote:
> Ah, I missed this. Looks good, thanks.
>
>
> On 25 February 2014 16:31, Ben Pfaff wrote:
>
> >
Ah, I missed this. Looks good, thanks.
On 25 February 2014 16:31, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> tests/ofproto-dpif.at |8
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/tests/ofproto-dpif.at b/tests/ofproto-dpif.at
> index b64ab41..97953d
Looks good.
On 25 February 2014 16:31, Ben Pfaff wrote:
> revalidator_sweep__() does not provide udumps, so push_dump_ops() can't
> look them up.
>
> This is my fault: I introduced it during review.
>
> Signed-off-by: Ben Pfaff
> ---
> ofproto/ofproto-dpif-upcall.c | 15 ++-
> 1
I'm looking now.
On 25 February 2014 16:31, Ben Pfaff wrote:
> On Tue, Feb 25, 2014 at 04:18:38PM -0800, Ben Pfaff wrote:
> > On Tue, Feb 25, 2014 at 04:17:12PM -0800, Joe Stringer wrote:
> > > On 25 February 2014 16:02, Ben Pfaff wrote:
> > >
> > > > On Tue, Feb 11, 2014 at 01:55:36PM -0800,
On Tue, Feb 25, 2014 at 04:18:38PM -0800, Ben Pfaff wrote:
> On Tue, Feb 25, 2014 at 04:17:12PM -0800, Joe Stringer wrote:
> > On 25 February 2014 16:02, Ben Pfaff wrote:
> >
> > > On Tue, Feb 11, 2014 at 01:55:36PM -0800, Joe Stringer wrote:
> > > > Previously, we would delete all ukeys when cha
Signed-off-by: Ben Pfaff
---
tests/ofproto-dpif.at |8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/tests/ofproto-dpif.at b/tests/ofproto-dpif.at
index b64ab41..97953d0 100644
--- a/tests/ofproto-dpif.at
+++ b/tests/ofproto-dpif.at
@@ -2793,9 +2793,9 @@ done
ovs-app
revalidator_sweep__() does not provide udumps, so push_dump_ops() can't
look them up.
This is my fault: I introduced it during review.
Signed-off-by: Ben Pfaff
---
ofproto/ofproto-dpif-upcall.c | 15 ++-
1 file changed, 10 insertions(+), 5 deletions(-)
diff --git a/ofproto/ofprot
On Tue, Feb 25, 2014 at 04:17:12PM -0800, Joe Stringer wrote:
> On 25 February 2014 16:02, Ben Pfaff wrote:
>
> > On Tue, Feb 11, 2014 at 01:55:36PM -0800, Joe Stringer wrote:
> > > Previously, we would delete all ukeys when changing the number of
> > > threads, but leave all flows in the datapat
On 25 February 2014 16:02, Ben Pfaff wrote:
> On Tue, Feb 11, 2014 at 01:55:36PM -0800, Joe Stringer wrote:
> > Previously, we would delete all ukeys when changing the number of
> > threads, but leave all flows in the datapath. This would cause
> > double-counting of stats for any flows that rema
Kmindg G added you to his circles and invited you to join Google+.
Click here to get started with Google+:
https://plus.google.com/_/notifications/emlink?emr=15965197021687490520&emid=COjRpKfB6LwCFciJRAodfmEAAA&path=%2F%3Fgpinv%3DAMIXal-yXYuldZBLnxRf9Fz6Z4QWm2EhPwUmrr4w-ZHtolPMtQGd1VPa5UwaGoAf9
Looks good.
On 25 February 2014 15:53, Ben Pfaff wrote:
> On Tue, Feb 11, 2014 at 01:55:35PM -0800, Joe Stringer wrote:
> > Previously, we would clean up the ukeys whose flow was not seen in the
> > most recent dump, while leaving the flow in the datapath. In the
> > unlikely case that the data
Ah, great. The original refactoring was mostly split out from the patchset
to get rid of the flow_dumper thread, which made it easier to split these
statistics fixes out. I hadn't realised there was a bit more that we could
do.
On 25 February 2014 15:51, Ben Pfaff wrote:
> On Tue, Feb 25, 2014
Yes, I moved it out of the function in the next patch.
On Tue, Feb 25, 2014 at 04:06:11PM -0800, Joe Stringer wrote:
> I suspect you may have found that the later patches make use of this
> structure in more locations?
>
> Either way, fine by me.
>
>
> On 25 February 2014 15:16, Ben Pfaff wrot
I suspect you may have found that the later patches make use of this
structure in more locations?
Either way, fine by me.
On 25 February 2014 15:16, Ben Pfaff wrote:
> On Tue, Feb 11, 2014 at 01:55:33PM -0800, Joe Stringer wrote:
> > It is possible for a datapath to dump the same flow twice, f
Yes, I was really satisfied to see that, because I'd spent some time trying to
debug that problem myself. The issue is pretty subtle.
On Tue, Feb 25, 2014 at 4:04 PM, Joe Stringer wrote:
> Thanks! Glad to hear that it fixes the intermittent failures.
>
>
> On 25 February 2014 14:58, Ben Pfaff w
Thanks! Glad to hear that it fixes the intermittent failures.
On 25 February 2014 14:58, Ben Pfaff wrote:
> On Tue, Feb 11, 2014 at 01:55:32PM -0800, Joe Stringer wrote:
> > Basic test to check that the datapath reports the correct number of
> > packets seen, after a delay of 1 second.
> >
> >
On Tue, Feb 11, 2014 at 01:55:36PM -0800, Joe Stringer wrote:
> Previously, we would delete all ukeys when changing the number of
> threads, but leave all flows in the datapath. This would cause
> double-counting of stats for any flows that remain in the datapath. This
> patch fixes the issue by en
On Tue, Feb 25, 2014 at 09:35:46AM -0800, Joe Stringer wrote:
> On 25 February 2014 00:01, Zhou, Han wrote:
>
> > Hi Joe,
> >
> > On Tue, 2014-02-11 at 13:55 -0800, Joe Stringer wrote:
> > > Previously, we would delete all ukeys when changing the number of
> > > threads, but leave all flows in th
On Tue, Feb 11, 2014 at 01:55:35PM -0800, Joe Stringer wrote:
> Previously, we would clean up the ukeys whose flow was not seen in the
> most recent dump, while leaving the flow in the datapath. In the
> unlikely case that the datapath fails to dump a flow that still exists
> in the datapath, this
On Tue, Feb 25, 2014 at 03:41:36PM -0800, Ben Pfaff wrote:
> On Tue, Feb 25, 2014 at 03:25:35PM -0800, Ben Pfaff wrote:
> > On Tue, Feb 11, 2014 at 01:55:34PM -0800, Joe Stringer wrote:
> > > This splits out functions for re-use by later patches, and compacts the
> > > udump revalidation code.
> >
Hi Guys,
This is version 2 for supporting NSH [1] on vxlan. The changes since
last patch set are few and mainly just related to the review earlier.
datapath/datapath.c|2 +
datapath/flow.h|7 +-
datapath/flow_netlink.c
From: pritesh
Support for nsh service index (nsi) is added, mainly incoming
nsi in a flow can be matched and appropriate action can be
taken on the flow based on it.
Signed-off-by: Pritesh Kothari
diff --git a/lib/flow.c b/lib/flow.c
index 28e96ec..e1b7d74 100644
--- a/lib/flow.c
+++ b/lib/flo
Support for setting nsi using an action namely set_nsi.
It works similar to set_tunnel in vxlan/gre tunnel and
can be used to set the outgoing nsh service index (nsi).
Also NXM_NX_NSI is defined here which enables matching
NSIs.
Signed-off-by: Pritesh Kothari
diff --git a/include/openflow/nicira
From: pritesh
Fix for the current test cases which fail due to the nsh changes.
Signed-off-by: Pritesh Kothari
diff --git a/tests/bfd.at b/tests/bfd.at
index 3723d60..23dbf87 100644
--- a/tests/bfd.at
+++ b/tests/bfd.at
@@ -229,7 +229,7 @@ OVS_VSWITCHD_START([add-port br0 p1 -- set Interface p
NSH service path (nsp) can be set/unset while creating the port
as well as nsp can be matched on incoming packets.
Signed-off-by: Pritesh Kothari
diff --git a/lib/flow.c b/lib/flow.c
index e7fe4d3..28e96ec 100644
--- a/lib/flow.c
+++ b/lib/flow.c
@@ -563,6 +563,8 @@ flow_tun_flag_to_string(uint3
Here, datapath support for setting nsh service index (nsi) is added.
nsi can now be set on outgoing packet depending on port settings or
set_nsi action.
Signed-off-by: Pritesh Kothari
diff --git a/datapath/datapath.c b/datapath/datapath.c
index b96ad1e..46a4f28 100644
--- a/datapath/datapath.c
+
Support for setting nsp using an action namely set_nsp.
It works similar to set_tunnel in vxlan/gre tunnel and
can be used to set the outgoing nsh service path id.
Also NXM_NX_NSP is defined which enables matching NSPs.
Signed-off-by: Pritesh Kothari
diff --git a/include/openflow/nicira-ext.h b/
This patch adds support for Network Service Headers (nsh) over VXLAN
as mentioned in [1]. Here changes are made to datapath to add nsh
headers whenever a vxlan port with destination port as 6633 (which is
IANA allocated port for nsh over vxlan) is created.
[1] http://tools.ietf.org/html/draft-quin
On Tue, Feb 25, 2014 at 03:25:35PM -0800, Ben Pfaff wrote:
> On Tue, Feb 11, 2014 at 01:55:34PM -0800, Joe Stringer wrote:
> > This splits out functions for re-use by later patches, and compacts the
> > udump revalidation code.
> >
> > Co-authored-by: Ethan Jackson
> > Signed-off-by: Joe Stringer
On Tue, Feb 11, 2014 at 01:55:34PM -0800, Joe Stringer wrote:
> This splits out functions for re-use by later patches, and compacts the
> udump revalidation code.
>
> Co-authored-by: Ethan Jackson
> Signed-off-by: Joe Stringer
I have this queued up to apply, thanks!
On Tue, Feb 11, 2014 at 01:55:33PM -0800, Joe Stringer wrote:
> It is possible for a datapath to dump the same flow twice, for instance
> if the flow is the last in a batch of flows to be dumped, then a new
> flow is inserted into the same bucket before the flow dumper fetches
> another batch.
>
>
On Mon, Feb 24, 2014 at 04:11:23PM +0100, Martino Fornasa wrote:
> The attached patch is an implementation of the Rapid Spanning Tree Protocol
> (RSTP, IEEE 802.1D-2004) on Open vSwitch.
>
> We have now performed some more in-depth tests on the implementation
> (you can find the test documentation
On Tue, Feb 11, 2014 at 01:55:32PM -0800, Joe Stringer wrote:
> Basic test to check that the datapath reports the correct number of
> packets seen, after a delay of 1 second.
>
> Signed-off-by: Joe Stringer
This series appear to fix the intermittent failure of test 683
"ofproto-dpif - controller
On Tue, Feb 25, 2014 at 9:45 AM, Linda Sun wrote:
> Sounds good to me.
Alright. I applied this on master.
>
> Linda
>
>
>
>> On Feb 24, 2014, at 8:34 PM, Gurucharan Shetty wrote:
>>
>>> On Wed, Feb 12, 2014 at 1:50 PM, Linda Sun wrote:
>>> Use send/recv for socket stream instead of read/write.
Hello,
I'm currently hitting a null pointer dereference and kernel panic that
seems to be in ovs. The problem is sporadic. I have one production machine
that's hit it four times in the past 24hrs, and one lab machine that I
can't get to hit it at all.
We rebuilt openvswitch with debugging symbols
If the .git directory is there, and "git --version" works, but "git
ls-files" fails to print any file names (e.g. perhaps because .git is a
symlink to an inaccessible directory) then the "grep" command in the
thread-safety-check rule would hang forever because it had an empty list
of files and ther
On 25 February 2014 00:01, Zhou, Han wrote:
> Hi Joe,
>
> On Tue, 2014-02-11 at 13:55 -0800, Joe Stringer wrote:
> > Previously, we would delete all ukeys when changing the number of
> > threads, but leave all flows in the datapath. This would cause
> > double-counting of stats for any flows that
Sounds good to me.
Linda
> On Feb 24, 2014, at 8:34 PM, Gurucharan Shetty wrote:
>
>> On Wed, Feb 12, 2014 at 1:50 PM, Linda Sun wrote:
>> Use send/recv for socket stream instead of read/write.
>> Use event handle for polling on socket stream.
>> Check windows specific return code.
>>
>> S
Thanks, applied to master, branch-2.0, branch-2.1.
On Mon, Feb 24, 2014 at 12:25:41PM -0800, Alex Wang wrote:
> Acked-by: Alex Wang
>
>
>
> On Mon, Feb 24, 2014 at 11:10 AM, Ben Pfaff wrote:
>
> > The following scenario can occur:
> >
> >1. Handler thread grabs a pointer to an ofproto in
Hi Edwin,
As a leading database provider, I was inquisitive to check if you would be
interested in owning a list of companies using VMware that can be used for
your marketing campaigns? These lists can be customized based on
application, geography, titles and more.
Some of the other Virtual
Hi Joe,
On Tue, 2014-02-11 at 13:55 -0800, Joe Stringer wrote:
> Previously, we would delete all ukeys when changing the number of
> threads, but leave all flows in the datapath. This would cause
> double-counting of stats for any flows that remain in the datapath. This
> patch fixes the issue by
48 matches
Mail list logo
