On Tue, Mar 31, 2015 at 11:23:32PM +0100, Mark Kavanagh wrote:
DPDK 2.0 contains an updated implementation of rte_memcopy, which
leverages SSE and AVX instrinsics. In versions of GCC = 4.8.2, it
has been observed that the relevant instrinsics are not defined in
gcc headers, resulting in
On 03/31/15 at 09:17pm, Ben Pfaff wrote:
Signed-off-by: Ben Pfaff b...@nicira.com
---
I'd like to order some OVN stickers for distribution at OpenStack and by
mail, so here's my OVN logo proposal. Also available at:
http://benpfaff.org/~blp/ovn.png
I got a Business Proposal for you, Email me for details via::
mrs.kevinc...@torba.com
Best Regards,
Mrs. Kevin Chen.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 04/01/15 at 11:52am, Thomas Graf wrote:
On 03/31/15 at 09:17pm, Ben Pfaff wrote:
Signed-off-by: Ben Pfaff b...@nicira.com
---
I'd like to order some OVN stickers for distribution at OpenStack and by
mail, so here's my OVN logo proposal. Also available at:
On 04/01/15 at 07:45am, Kavanagh, Mark B wrote:
What's wrong with setting CFLAGS on the configure or make command
line? This is the standard way to do this with Automake and Autoconf.
Sure. However, setting CFLAGS on the command line overwrites any values
CFLAGS has attained via the
Added a new IOCTL in order to retrieve the PID from the kernel datapath.
The new method uses a direct and cleaner way, as opposed to the old way
of using a Netlink transaction, avoiding the unnecessary overhead.
Signed-off-by: Sorin Vinturis svintu...@cloudbasesolutions.com
Reported-by: Alin
On Wed, Apr 1, 2015 at 5:03 AM, Thomas Graf tg...@noironetworks.com wrote:
On 04/01/15 at 11:52am, Thomas Graf wrote:
On 03/31/15 at 09:17pm, Ben Pfaff wrote:
Signed-off-by: Ben Pfaff b...@nicira.com
---
I'd like to order some OVN stickers for distribution at OpenStack and
by
On Wed, Apr 01, 2015 at 08:15:52AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 5:03 AM, Thomas Graf tg...@noironetworks.com wrote:
On 04/01/15 at 11:52am, Thomas Graf wrote:
On 03/31/15 at 09:17pm, Ben Pfaff wrote:
Signed-off-by: Ben Pfaff b...@nicira.com
---
I'd like to
The counters use x86 TSC if available (currently only with DPDK). They
will be exposed by subsequents commits
Signed-off-by: Daniele Di Proietto diproiet...@vmware.com
---
lib/dpif-netdev.c | 70 ++-
1 file changed, 69 insertions(+), 1
These commands can be used to get packets and cycles counters on a pmd
thread basis. They're useful to get a clearer picture about the
performance of the userspace datapath.
They export these pieces of information:
- A (per-thread) view of the caches hit rate. Hits in the exact match
cache
This init function is called when the dpif class is registered. It will
be used by following commits
Signed-off-by: Daniele Di Proietto diproiet...@vmware.com
---
lib/dpif-netdev.c | 1 +
lib/dpif-netlink.c | 1 +
lib/dpif-provider.h | 8
lib/dpif.c | 8
4 files
On Wed, Apr 01, 2015 at 11:34:39AM +0200, Thomas Graf wrote:
On 04/01/15 at 07:45am, Kavanagh, Mark B wrote:
Agreed; given the precedent set by OVS_ENABLE_WERROR, I figured that
this was the convention.
The only reason this still exists is for backwards compatibility
to not break existing
On Wed, Apr 01, 2015 at 07:45:10AM +, Kavanagh, Mark B wrote:
On Tue, Mar 31, 2015 at 11:23:32PM +0100, Mark Kavanagh wrote:
DPDK 2.0 contains an updated implementation of rte_memcopy, which
leverages SSE and AVX instrinsics. In versions of GCC = 4.8.2, it
has been observed that the
From: Billy O'Mahony billy.o.mah...@intel.com
When creating bonded ports using DPDK-enabled interfaces it is required that
the interface type is set explicitly. This is not required when using 'system'
type interfaces. This patch adds documentation to this effect.
I would be interested in
From: Billy O'Mahony billy.o.mah...@intel.com
Unlike system interfaces, DPDK enabled interfaces must have their interface
type explicitly set when used to create bonded ports. Mention this at the
relevant points in the documentation.
Signed-off-by: Billy O'Mahony billy.o.mah...@intel.com
---
On Wed, Apr 01, 2015 at 11:12:39AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 10:39 AM, Ben Pfaff b...@nicira.com wrote:
On Wed, Apr 01, 2015 at 08:15:52AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 5:03 AM, Thomas Graf tg...@noironetworks.com
wrote:
On 04/01/15 at
A read operation from a non atomic shared value (without external
locking) can return incorrect values. Using the atomic semantics
prevents this from happening.
However:
* No memory barriers are used. We don't need that kind of consistency
for statistics (we use relaxed operations).
* The
On Wed, Apr 1, 2015 at 10:39 AM, Ben Pfaff b...@nicira.com wrote:
On Wed, Apr 01, 2015 at 08:15:52AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 5:03 AM, Thomas Graf tg...@noironetworks.com
wrote:
On 04/01/15 at 11:52am, Thomas Graf wrote:
On 03/31/15 at 09:17pm, Ben Pfaff
We used to count exact match cache hits and masked classifier hits
together. This commit splits the DP_STAT_HIT counter into two.
This change will be used by future commits.
Signed-off-by: Daniele Di Proietto diproiet...@vmware.com
---
lib/dpif-netdev.c | 16 ++--
1 file changed, 10
I think a better title for this patch would be something like
jsonrpc-server: Split monitors into database back end and JSON-RPC
front end.
On Thu, Mar 19, 2015 at 12:08:17AM -0700, Andy Zhou wrote:
jsonrpc-server.c has two main functions. One deals with handling the
jsonrpc connections, the
On 30 Mar 2015, at 22:41, Ethan Jackson et...@nicira.com wrote:
You're not going to like my review of this patch =)
Anything to try to make the code better :-)
So the code is very high quality, though I haven't read it super
closely yet. My main question, is if this is really the right
Since flow statistics are thread local and updated without any lock,
it is not correct to do a memset from another thread.
This commit simply removes the support for the flag. It is not needed
by ofproto-dpif, it is only exposed by dpctl commands.
Signed-off-by: Daniele Di Proietto
On Apr 1, 2015, at 9:17 AM, Ben Pfaff b...@nicira.com wrote:
On Wed, Apr 01, 2015 at 11:12:39AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 10:39 AM, Ben Pfaff b...@nicira.com wrote:
How about like this:
http://benpfaff.org/~blp/ovn4.png
On Thu, Mar 19, 2015 at 12:08:19AM -0700, Andy Zhou wrote:
Change ovsdb_jsonrpc_parse_monitor_request() to make
ovsdb_monitor_table an opaque object.
Signed-off-by: Andy Zhou az...@nicira.com
I'm not sure that this is the best way to check for duplicates if
ovsdb_monitor is supposed to be
The goal of this series is to add an appctl command to retrieve detailed
performance statistics from the pmd thread execution. These statistics
include exact match cache and masked classifier hits and rough cycles counters.
This series also fixes some concurrency issues with statistics in the
Since statistics updates might require locking (in future commits)
grouping them will reduce the locking overhead.
Signed-off-by: Daniele Di Proietto diproiet...@vmware.com
---
lib/dpif-netdev.c | 11 ++-
1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/lib/dpif-netdev.c
On Tue, Mar 31, 2015 at 05:00:44PM -0700, Ben Pfaff wrote:
On Tue, Mar 31, 2015 at 04:03:04PM -0400, Russell Bryant wrote:
Signed-off-by: Russell Bryant rbry...@redhat.com
---
This is very early, but I wanted to get feedback on the general approach
used to
detect changes in the ovn
The original message was received at Thu, 2 Apr 2015 08:35:40 +0800 from
yuriko.or.jp [35.93.76.97]
- The following addresses had permanent fatal errors -
dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
Yes, I like that better, thanks!
On Wed, Apr 01, 2015 at 04:59:42PM -0700, Alex Wang wrote:
How about this?
+} else if (!ovsdb_idl_has_lock(idl)
+ || !ovsdb_idl_has_ever_connected(idl)) {
+/* Returns if not holding the lock or not done retrieving db
+ *
I've been looking at filling in the sFlow structures to report on tunnel
encap and other transformations. sFlow sampling is best done on ingress
only, so I can't use the egress-sampling action that the IPFIX
implementation uses to get the tunnel info. So how should I look up the
list of
On Wed, Apr 01, 2015 at 01:40:07PM -0400, Russell Bryant wrote:
A new transaction is allocated before executing the command. If the
result from committing the transaction is TRY_AGAIN, the code leaked
the allocated transaction since it creates a new one when it comes
back around to retry.
How about this?
+} else if (!ovsdb_idl_has_lock(idl)
+ || !ovsdb_idl_has_ever_connected(idl)) {
+/* Returns if not holding the lock or not done retrieving db
+ * contents. */
return;
}
On Wed, Apr 1, 2015 at 4:55 PM, Ben Pfaff b...@nicira.com
hi Sorin,
Thanks for the patch. It is great that you are addressing many issues that have
been pending for sometime.
I don’t know if there’s a whole lot of technical advantage to using an separate
ioctl v/s using a netlink command, but I can see that it makes the
get_sock_pid_from_kernel()
On Thu, Mar 19, 2015 at 12:08:16AM -0700, Andy Zhou wrote:
This patch set implements two ideas improve ovsdb connection scaling:
* Allow multiple jsonrpc connection to share a single ovsdb monitor, thus
maintaining a single set of changes.
* With in the same monitor, cache the json
On Wed, Apr 01, 2015 at 04:52:24PM -0700, Alex Wang wrote:
During upgrade of ovs-vswitchd, we do not want to recreate the kernel
interfaces. Especially when IP address is assigned to the internal port,
the recreation will cause the lost of connection. Therefore, ovs-vswitchd
should read
On Thu, Mar 19, 2015 at 12:08:42AM -0700, Andy Zhou wrote:
Although multiple jsonrpc monitor can share the same ovsdb monitor, each
change still needs to transalted into json object from scratch. This
can be wastful is mutiple jsonrpc monitors are interested in the
same changes.
Json chche
Thx, applied to master~
On Wed, Apr 1, 2015 at 5:02 PM, Ben Pfaff b...@nicira.com wrote:
Yes, I like that better, thanks!
On Wed, Apr 01, 2015 at 04:59:42PM -0700, Alex Wang wrote:
How about this?
+} else if (!ovsdb_idl_has_lock(idl)
+ ||
Hello,
I am working on PF/VF driver for PCI-E based SR-IOV capable card. This device
has got few cpus and ram and is capable of running linux and any linux-userland
application on the device it self. Register layout (for datapath like TX/RX
ring, stats etc.) between host and device is defined
Seems likely.
For new kernels, make sure you're building master.
On Wed, Apr 01, 2015 at 04:46:54PM -0700, Sabyasachi Sengupta wrote:
ovs from master branch off github does not build in C7.1. There are
quite a few errors in datapath directory, was hoping if a patch was
already available.
During upgrade of ovs-vswitchd, we do not want to recreate the kernel
interfaces. Especially when IP address is assigned to the internal port,
the recreation will cause the lost of connection. Therefore, ovs-vswitchd
should read current ovsdb content first and then reuse the existing kernel
Does it build? Then I imagine it's supported.
On Wed, Apr 01, 2015 at 01:44:41PM -0700, Sabyasachi Sengupta wrote:
Can anyone respond to this question? I can provide a patch for enabling the
3.10.229 kernel (backported from native Linux) if one is not already
available..
On Mon, 30 Mar
On Thu, Apr 02, 2015 at 12:41:53AM +0200, Thomas Graf wrote:
On 04/01/15 at 08:06am, Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 11:34:39AM +0200, Thomas Graf wrote:
On 04/01/15 at 07:45am, Kavanagh, Mark B wrote:
Agreed; given the precedent set by OVS_ENABLE_WERROR, I figured that
this
ovs from master branch off github does not build in C7.1. There are quite a
few errors in datapath directory, was hoping if a patch was already
available. Probably no one tried it yet?
---
Nuage Networks Business Unit, Alcatel-Lucent,
755 Ravendale Drive, Mountain View, CA, 94043
Mailstop:
These changes are just some minor changes I have made to the skeleton
since the version that has been merged. It adds the daemon related
options, updates the table/columns we monitor from the OVN db, and
some other minor tweaks.
Signed-off-by: Russell Bryant rbry...@redhat.com
---
ovn/ovn-nbd.c
When the state of the chassis column in the Bindings table changes for
any row, ovn-nbd will notice and trigger recalculating the 'up' state
for all logical ports.
This can be tesed manually by starting up ovs-sandbox with ovn support
enabled, running ovn-nbd, and then running the following
On 04/01/2015 07:50 PM, Ben Pfaff wrote:
On Tue, Mar 31, 2015 at 05:00:44PM -0700, Ben Pfaff wrote:
On Tue, Mar 31, 2015 at 04:03:04PM -0400, Russell Bryant wrote:
Signed-off-by: Russell Bryant rbry...@redhat.com
---
This is very early, but I wanted to get feedback on the general approach
On Apr 1, 2015, at 1:24 PM, Ben Pfaff b...@nicira.com wrote:
On Thu, Mar 19, 2015 at 12:08:33AM -0700, Andy Zhou wrote:
jsonrpc_monitor_compse_update() seems fit better than
jsonrpc_monitor_compose_table_update(), since it composes changes
from all tables.
Signed-off-by: Andy Zhou
On Thu, Apr 02, 2015 at 12:35:21AM +0200, Thomas Graf wrote:
On 04/01/15 at 09:17am, Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 11:12:39AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 10:39 AM, Ben Pfaff b...@nicira.com wrote:
How about like this:
I guess I feel like it's not necessary for the logo to emphasize this.
On Wed, Apr 01, 2015 at 10:16:04AM -0700, Steven Noble wrote:
I am terrible at art but maybe something including a north south arrow
possibly incorporated into the N? Switch is flat, Network is not imho.
On Wed, Apr 01, 2015 at 10:55:44PM -0700, Justin Pettit wrote:
On Apr 1, 2015, at 1:24 PM, Ben Pfaff b...@nicira.com wrote:
On Thu, Mar 19, 2015 at 12:08:33AM -0700, Andy Zhou wrote:
jsonrpc_monitor_compse_update() seems fit better than
jsonrpc_monitor_compose_table_update(), since it
On Wed, Apr 01, 2015 at 09:21:01AM -0700, Justin Pettit wrote:
On Apr 1, 2015, at 9:17 AM, Ben Pfaff b...@nicira.com wrote:
On Wed, Apr 01, 2015 at 11:12:39AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 10:39 AM, Ben Pfaff b...@nicira.com wrote:
How about like this:
I am terrible at art but maybe something including a north south arrow
possibly incorporated into the N? Switch is flat, Network is not imho.
https://www.sonn.com/ovn6.svg
Ben Pfaff wrote:
That's part of the OVS logo embedded in the OVN logo, see
http://openvswitch.org/assets/vswitch.png.
Believe my word, this sale is your last chance, for crying out loud!
According to a recent survey American men had fewer health tests and
investigations this year.
http://x.co/8JE9Y
___
dev mailing list
dev@openvswitch.org
On Wed, Apr 01, 2015 at 09:50:41AM -0700, Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 09:21:01AM -0700, Justin Pettit wrote:
On Apr 1, 2015, at 9:17 AM, Ben Pfaff b...@nicira.com wrote:
On Wed, Apr 01, 2015 at 11:12:39AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 10:39 AM, Ben
That's part of the OVS logo embedded in the OVN logo, see
http://openvswitch.org/assets/vswitch.png.
On Wed, Apr 01, 2015 at 09:56:49AM -0700, Steven Noble wrote:
Is there a reason you only have east-west arrows?
Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 09:50:41AM -0700, Ben Pfaff wrote:
On
A new transaction is allocated before executing the command. If the
result from committing the transaction is TRY_AGAIN, the code leaked
the allocated transaction since it creates a new one when it comes
back around to retry. The old transaction is now destroyed before
continuing to allow the
On Wed, Apr 01, 2015 at 12:16:33PM -0700, Ben Pfaff wrote:
On Thu, Mar 19, 2015 at 12:08:25AM -0700, Andy Zhou wrote:
Added new files ovsdb-monitor.[ch] for monitor backend functions.
There is no functional changes.
Signed-off-by: Andy Zhou az...@nicira.com
Acked-by: Ben Pfaff
On Thu, Mar 19, 2015 at 12:08:27AM -0700, Andy Zhou wrote:
Refactoring ovsdb_monitor_get_initial() to not generate JSON object.
It only collect changes within the ovsdb_monitor().
ovsdb_jsonrpc_monitor_compose_table_update() is then used to generate
JSON object.
This change will also make
On Thu, Mar 19, 2015 at 12:08:25AM -0700, Andy Zhou wrote:
Added new files ovsdb-monitor.[ch] for monitor backend functions.
There is no functional changes.
Signed-off-by: Andy Zhou az...@nicira.com
Acked-by: Ben Pfaff b...@nicira.com
___
dev
On Thu, Mar 19, 2015 at 12:08:22AM -0700, Andy Zhou wrote:
split out per monitoring needs_flush() into
ovsdb_monitor_needs_flush().
Signed-off-by: Andy Zhou az...@nicira.com
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
On Thu, Mar 19, 2015 at 12:08:23AM -0700, Andy Zhou wrote:
rename ovsdb_jsonrpc_monitor_get_initial() to
ovsdb_monitor_get_initial()
Signed-off-by: Andy Zhou az...@nicira.com
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
On Thu, Mar 19, 2015 at 12:08:24AM -0700, Andy Zhou wrote:
Add ovsdb_minitor_destory() function to properly cleanup ovsdb_monitor.
s/minitor/monitor/
It is also responsible for unhook from the replica chain.
The replica destroy callback is now called
ovsdb_monitor_destroy_callback()
On Thu, Mar 19, 2015 at 12:08:20AM -0700, Andy Zhou wrote:
To hide ovsdb_monitor_table object from ovsdb_jsonrpc serve.
Signed-off-by: Andy Zhou az...@nicira.com
It looks like there should be indentation adjustment here:
static void
-ovsdb_monitor_set_select(struct ovsdb_monitor_table *mt,
Is there a reason you only have east-west arrows?
Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 09:50:41AM -0700, Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 09:21:01AM -0700, Justin Pettit wrote:
On Apr 1, 2015, at 9:17 AM, Ben Pfaffb...@nicira.com wrote:
On Wed, Apr 01, 2015 at 11:12:39AM -0500,
On Thu, Mar 19, 2015 at 12:08:34AM -0700, Andy Zhou wrote:
Currently, each monitor table contains a single hmap 'changes' to
track updates. This patch introduces a new data structure
'ovsdb_monitor_changes' that stores the updates 'rows' tagged by
its first commit transaction id. Each
On Wed, Apr 01, 2015 at 02:15:12PM -0700, Ben Pfaff wrote:
On Thu, Mar 19, 2015 at 12:08:34AM -0700, Andy Zhou wrote:
Currently, each monitor table contains a single hmap 'changes' to
track updates. This patch introduces a new data structure
'ovsdb_monitor_changes' that stores the updates
On Thu, Mar 19, 2015 at 12:08:39AM -0700, Andy Zhou wrote:
Add ovsdb_monitor_add_jsonrpc_monitor(). This change will allow
ovsdb_monitor to be reference counted.
Signed-off-by: Andy Zhou az...@nicira.com
Acked-by: Ben Pfaff b...@nicira.com
'void' should be on a line by its own here:
+void
On Thu, Mar 19, 2015 at 12:08:21AM -0700, Andy Zhou wrote:
Now it simply calls ovsdb_monitor_compose_table_update(), which
is actually creates the json object.
Signed-off-by: Andy Zhou az...@nicira.com
Acked-by: Ben Pfaff b...@nicira.com
___
dev
On Wed, Apr 01, 2015 at 05:21:08PM +0100, Daniele Di Proietto wrote:
Minor issue: the compilers I’ve tried (GCC 4.9, 4.8 clang 3.5 and sparse
0.4.5) do not report a warning if I try to access atomic variables without
atomic function. Maybe someone can confirm this?
Yeah, those versions of GCC
On Thu, Mar 19, 2015 at 12:08:28AM -0700, Andy Zhou wrote:
Currently, each ovsdb-monitor points to a single jsonrpc_monitor object.
This means there is 1:1 relationship between them.
In case multiple jsonrpc-monitors needs to monitor the same tables and
the columns within them, then can
On Thu, Mar 19, 2015 at 12:08:30AM -0700, Andy Zhou wrote:
With N:1 mappings, multiple jsonrpc server may be servicing the rpc
connection at a different pace. ovsdb-monitor thus needs to maintain
different change sets, depends on connection speed of each rpc
connections. Connections servicing
Can anyone respond to this question? I can provide a patch for enabling the
3.10.229 kernel (backported from native Linux) if one is not already
available..
On Mon, 30 Mar 2015, Sabyasachi Sengupta wrote:
Hi,
Can anyone please let me know if Centos 7.1 (3.10.0-229) kernel supported in
On Thu, Mar 19, 2015 at 12:08:33AM -0700, Andy Zhou wrote:
jsonrpc_monitor_compse_update() seems fit better than
jsonrpc_monitor_compose_table_update(), since it composes changes
from all tables.
Signed-off-by: Andy Zhou az...@nicira.com
I think it was named after the table-updates object
On Thu, Mar 19, 2015 at 12:08:34AM -0700, Andy Zhou wrote:
Currently, each monitor table contains a single hmap 'changes' to
track updates. This patch introduces a new data structure
'ovsdb_monitor_changes' that stores the updates 'rows' tagged by
its first commit transaction id. Each
On 04/01/15 at 09:17am, Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 11:12:39AM -0500, Kyle Mestery wrote:
On Wed, Apr 1, 2015 at 10:39 AM, Ben Pfaff b...@nicira.com wrote:
How about like this:
http://benpfaff.org/~blp/ovn4.png
http://benpfaff.org/~blp/ovn4.svg
The
On 04/01/15 at 08:06am, Ben Pfaff wrote:
On Wed, Apr 01, 2015 at 11:34:39AM +0200, Thomas Graf wrote:
On 04/01/15 at 07:45am, Kavanagh, Mark B wrote:
Agreed; given the precedent set by OVS_ENABLE_WERROR, I figured that
this was the convention.
The only reason this still exists is for
On Thu, Mar 19, 2015 at 12:08:41AM -0700, Andy Zhou wrote:
Store ovsdb monitor in global hmap. A newly created ovsdb monitor
object will first search the global hmap for a possible match. If
one is found, the existing ovsdb monitor is used instead.
With this patch, jsonrpc monitor and ovsdb
77 matches
Mail list logo
