On Thu, 10 Aug 2023 14:13:29 -0400 Aaron Conole wrote:
> I think they can be resolved in the same way the mac80211 drops are
> resolved by using (__force u32) to pass the reason argument.
Yup, preferably by creating a helper which takes enum ovs_drop_reason
and does the forcing, rather than
Only three families use info->userhdr today and going forward
we discourage using fixed headers in new families.
So having the pointer to user header in struct genl_info
is an overkill. Compute the header pointer at runtime.
Reviewed-by: Johannes Berg
Reviewed-by: Jiri Pirko
Signed-off-by:
On 8/4/23 19:25, Colin Watson wrote:
> tbl defaults to expecting table entries to be separated by tab
> characters. However, commit 5a0e4aec1af5cf7741c490bce704577e51e536b9
> converted these to spaces and inadvertently broke the rendering. Use
> semicolons as separators instead; these are less
On 8/4/23 19:25, Colin Watson wrote:
> groff 1.23.0 has difficulty with hyphenating a number of the entries in
> the name column of these tables when rendering in 80-column terminals.
> Setting a minimum width for this column gives it an easier time.
>
> Reported-by: Lucas Nussbaum
>
On 8/4/23 19:24, Colin Watson wrote:
> Using "NXM/OXM Support" makes these tables a little too wide to fit well
> when rendered in 80 columns, causing warnings from groff. There's
> already some abbreviation going on here (e.g. "RW?"), so "NXM/OXM?"
> seems acceptable.
>
> Reported-by: Lucas
On 8/4/23 19:24, Colin Watson wrote:
> This fixes a number of "table wider than line length minus indentation"
> warnings from tbl. The table cells are too narrow for centered text to
> look good, so left-align the contents of the text blocks.
>
> Reported-by: Lucas Nussbaum
> Reported-at:
Adrian Moreno writes:
> Create a new drop reason subsystem for openvswitch and add the first
> drop reason to represent last-action drops.
>
> Last-action drops happen when a flow has an empty action list or there
> is no action that consumes the packet (output, userspace, recirc, etc).
> It is
Adrian Moreno writes:
> Test explicit drops generate the right drop reason. Also, verify that
> the kernel rejects flows with actions following an explicit drop.
>
> Signed-off-by: Adrian Moreno
> ---
Acked-by: Aaron Conole
___
dev mailing list
On 8/10/23 18:15, Han Zhou wrote:
> Thanks Vladislav and Dumitru for reporting and fixing the issue.
> The impact of the issue is more than just the memory spikes in
> ovn-controller. More importantly, it incurs much higher load on SB DB
> because the conditions are flooded with all the localnet
On 8/10/23 17:34, Dumitru Ceara wrote:
> On 8/10/23 17:20, Han Zhou wrote:
>> On Thu, Aug 10, 2023 at 6:36 AM Dumitru Ceara wrote:
>>>
>>> On 8/10/23 15:34, Han Zhou wrote:
On Thu, Aug 10, 2023 at 2:29 AM Dumitru Ceara wrote:
>
> On 8/10/23 08:12, Ales Musil wrote:
>> On Wed,
Thanks Vladislav and Dumitru for reporting and fixing the issue.
The impact of the issue is more than just the memory spikes in
ovn-controller. More importantly, it incurs much higher load on SB DB
because the conditions are flooded with all the localnet ports regardless
of whether they belong to
Today the localnet ports are added to local_lports unconditionally.
This has several side-effects:
- When ovn-monitor-all is true, all localnet ports are added to
local_lports, including the ones on the non-local logical switches,
which is undesirable, although there is no direct consequences
On Thu, Aug 10, 2023 at 8:34 AM Dumitru Ceara wrote:
>
> On 8/10/23 17:20, Han Zhou wrote:
> > On Thu, Aug 10, 2023 at 6:36 AM Dumitru Ceara wrote:
> >>
> >> On 8/10/23 15:34, Han Zhou wrote:
> >>> On Thu, Aug 10, 2023 at 2:29 AM Dumitru Ceara
wrote:
>
> On 8/10/23 08:12, Ales Musil
On 8/10/23 17:20, Han Zhou wrote:
> On Thu, Aug 10, 2023 at 6:36 AM Dumitru Ceara wrote:
>>
>> On 8/10/23 15:34, Han Zhou wrote:
>>> On Thu, Aug 10, 2023 at 2:29 AM Dumitru Ceara wrote:
On 8/10/23 08:12, Ales Musil wrote:
> On Wed, Aug 9, 2023 at 5:13 PM Mark Michelson
>>> wrote:
On Thu, Aug 10, 2023 at 6:36 AM Dumitru Ceara wrote:
>
> On 8/10/23 15:34, Han Zhou wrote:
> > On Thu, Aug 10, 2023 at 2:29 AM Dumitru Ceara wrote:
> >>
> >> On 8/10/23 08:12, Ales Musil wrote:
> >>> On Wed, Aug 9, 2023 at 5:13 PM Mark Michelson
> > wrote:
> >>>
> Hi Ales,
>
> I
On Thu, Aug 10, 2023 at 7:51 PM Numan Siddique wrote:
>
> On Wed, Jul 26, 2023 at 3:58 PM Lorenzo Bianconi
> wrote:
> >
> > ovn supports creating remote logical ports. An user
> > can set requested-chassis option for a logical switch port
> > to the remote chassis and ovn-northd can bind it to
On Wed, Jul 26, 2023 at 3:58 PM Lorenzo Bianconi
wrote:
>
> ovn supports creating remote logical ports. An user
> can set requested-chassis option for a logical switch port
> to the remote chassis and ovn-northd can bind it to that chassis.
> This is required for OVN IC in ovn-k8s. Right now
Add config option called "northd-backoff-interval-ms" that allows
to delay northd engine runs capped by the config option.
When the config option is set to 0 or unspecified, the engine
will run without any restrictions. If the value >0 we will delay
northd engine run by the previous run time
On Wed, Aug 9, 2023 at 11:19 AM Simon Horman wrote:
>
> On Wed, Aug 09, 2023 at 06:42:21PM +0800, Dexia Li via dev wrote:
> > The netdev receiving packets will be encapsulated if output port
> > is tunnel type. Only support userspace vxlan or geneve tunnel tso
> > and inner csum offload.
On 8/10/23 15:34, Han Zhou wrote:
> On Thu, Aug 10, 2023 at 2:29 AM Dumitru Ceara wrote:
>>
>> On 8/10/23 08:12, Ales Musil wrote:
>>> On Wed, Aug 9, 2023 at 5:13 PM Mark Michelson
> wrote:
>>>
Hi Ales,
I have some high-level comments/questions about this patch.
>>>
>>> Hi
On Thu, Aug 10, 2023 at 2:29 AM Dumitru Ceara wrote:
>
> On 8/10/23 08:12, Ales Musil wrote:
> > On Wed, Aug 9, 2023 at 5:13 PM Mark Michelson
wrote:
> >
> >> Hi Ales,
> >>
> >> I have some high-level comments/questions about this patch.
> >>
> >
> > Hi Mark,
> >
>
> Hi Ales, Mark,
>
> > thank
On 10/08/2023 15:26, Ilya Maximets wrote:
> On 8/10/23 09:59, Roi Dayan wrote:
>> Hi,
>>
>> We noticed a possible memory leak in netdev_tc_init_flow_api()
>> as we allocate memory for meter_police_ids with id_pool_create()
>> but we never destroy it.
>
> Hi, Roi.
>
> I'd not consider this as
Add support for setting CT zone limit via ovs-vswitchd
database CT_Zone entry. The limit is propagated
into corresponding datapath. In order to keep
backward compatibility the dpctl/ct-set-limits command
can overwrite the database settings.
Signed-off-by: Ales Musil
---
ofproto/ofproto-dpif.c
It's similar to the processing we do for address sets. There's a bit
more mechanics involved due to the fact that we need to split NB port
groups per datapath.
We currently only partially implement incremental processing of
port_group changes in the lflow node. That is, we deal with the case
For now it's still a node recompute for every port group change. It
doesn't trigger northd recompute anymore though.
A follow up commit will add incremental processing of NB.Port_Group
changes.
Signed-off-by: Dumitru Ceara
---
lib/stopwatch-names.h|1 +
northd/en-lflow.c|4
No functional differences in this commit, just abstract out the
processing a bit.
Signed-off-by: Dumitru Ceara
---
northd/automake.mk |2
northd/en-port-group.c | 237 ++
northd/en-port-group.h | 63
northd/northd.c|
There's currently no need to store both 'ovn_ls_port_group'
and 'ovn_port_group_ls' so just use the first type.
Signed-off-by: Dumitru Ceara
---
northd/en-lflow.c |2
northd/northd.c | 286 +
northd/northd.h | 25 -
3 files
It's actually a generic wrapper and will be useful for upcoming commits.
This commit doesn't perform any functional changes but cleans up a bit
the implementation unifying the sorted_array cleanup. Before the change
the caller had to track whether it should free the internal 'arr' field
or not.
This series cleans up, factors out and then adds incremental processing
support to the code that processes port group updates in ovn-northd.
Some performance results are shared in the commit message of the last
patch in this series.
Dumitru Ceara (5):
ovn-util: Factor out struct
On 8/10/23 09:59, Roi Dayan wrote:
> Hi,
>
> We noticed a possible memory leak in netdev_tc_init_flow_api()
> as we allocate memory for meter_police_ids with id_pool_create()
> but we never destroy it.
Hi, Roi.
I'd not consider this as a leak. The pool is allocated once and
stored in a static
On 8/10/23 08:12, Ales Musil wrote:
> On Wed, Aug 9, 2023 at 5:13 PM Mark Michelson wrote:
>
>> Hi Ales,
>>
>> I have some high-level comments/questions about this patch.
>>
>
> Hi Mark,
>
Hi Ales, Mark,
> thank you for the review. See my answers inline below.
>
>
>> I have been privy to
Wed, Aug 09, 2023 at 08:26:41PM CEST, k...@kernel.org wrote:
>Only three families use info->userhdr and fixed headers
>are discouraged for new families. So remove the pointer
>from struct genl_info to save some space. Compute
>the header pointer at runtime. Saved space will be used
>for a family
Hi,
We noticed a possible memory leak in netdev_tc_init_flow_api()
as we allocate memory for meter_police_ids with id_pool_create()
but we never destroy it.
I added a code to refcount netdev_tc_init_flow_api and so in
netdev_tc_uninit_flow_api() to destroy it and noticed some behavior
about
On Wed, Aug 9, 2023 at 5:13 PM Mark Michelson wrote:
> Hi Ales,
>
> I have some high-level comments/questions about this patch.
>
Hi Mark,
thank you for the review. See my answers inline below.
> I have been privy to the conversations that led to this change. My
> understanding is that by
34 matches
Mail list logo