This comment-only patch could use a review.
On Wed, Apr 14, 2021 at 07:45:04PM -0700, Ben Pfaff wrote:
> Commit a14502a7c529 ("classifier: Retire partitions.") in 2015 removed
> OVS support for tags for the second time (!), so these comments about
> them should be removed t
This documentation-only patch could use a review.
On Wed, Apr 14, 2021 at 08:34:46PM -0700, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> lib/ovs-actions.xml | 288 +++-
> 1 file changed, 286 insertions(+), 2 deletions(-)
>
> di
On Fri, Apr 16, 2021 at 12:41:13PM +0200, Timothy Redaelli wrote:
> On Wed, 14 Apr 2021 13:40:20 -0700
> Ben Pfaff wrote:
>
> > The action headings were coming out all smashed together, like
> > "Theoutputaction". This fixes them so that they appear correct
On Thu, May 06, 2021 at 12:37:36AM +0200, Gaëtan Rivet wrote:
> On Wed, May 5, 2021, at 21:36, Ben Pfaff wrote:
> > On Wed, Apr 28, 2021 at 01:03:24AM +0200, Gaetan Rivet wrote:
> > > This series adds a compilation option that changes the behavior of the RCU
> > >
On Thu, May 06, 2021 at 12:34:00AM +0200, Ilya Maximets wrote:
> On 5/5/21 11:40 PM, Ben Pfaff wrote:
> > I think that this series has the details that one needs to understand it
> > at a high level, but I didn't see a high-level overview of its intended
> > u
On Sat, May 01, 2021 at 02:55:41AM +0200, Ilya Maximets wrote:
> Replication can be used to scale out read-only access to the database.
> But there are clients that are not read-only, but read-mostly.
> One of the main examples is ovn-controller that mostly monitors
> updates from the Southbound
On Wed, Apr 28, 2021 at 01:03:24AM +0200, Gaetan Rivet wrote:
> This series adds a compilation option that changes the behavior of the RCU
> module. Once enabled, RCU reclamation by user threads becomes blocking until
> the RCU threads has executed the scheduled callbacks.
It's a great idea to
must have fixed this or
> something.
>
> On 4/12/21 12:44 PM, Ben Pfaff wrote:
> > Hi Mark! This is a great question. I'm taking a few days off, and
> > Leonid is on paternity leave, so I'll take a look at this toward the end
> > of the week.
> >
> > That said,
Signed-off-by: Ben Pfaff
---
lib/ovs-actions.xml | 288 +++-
1 file changed, 286 insertions(+), 2 deletions(-)
diff --git a/lib/ovs-actions.xml b/lib/ovs-actions.xml
index a2778de4bcd6..de934a244de9 100644
--- a/lib/ovs-actions.xml
+++ b/lib/ovs
Commit a14502a7c529 ("classifier: Retire partitions.") in 2015 removed
OVS support for tags for the second time (!), so these comments about
them should be removed too.
Signed-off-by: Ben Pfaff
---
lib/mac-learning.c | 4
1 file changed, 4 deletions(-)
diff --git a/lib/mac-le
t;, and " action" each got
stripped down to "The", "output", "action" after processing. This
commit changes it so that stripping happens after concatenation, fixing
the problem.
Signed-off-by: Ben Pfaff
---
python/build/nroff.py | 7 +++
1 file changed, 3
On Fri, Apr 09, 2021 at 06:14:24PM +0200, Ilya Maximets wrote:
> On 4/9/21 6:01 PM, Mark Michelson wrote:
> > Hi guys,
> >
> > While developing a new feature, I found a defect in ovn-northd-ddlog when
> > HA_Chassis_Groups are used. I have attached a script that replicates the
> > issue.
> >
>
Hi Mark! This is a great question. I'm taking a few days off, and
Leonid is on paternity leave, so I'll take a look at this toward the end
of the week.
That said, one possibility that comes up sometimes is that northd is
sending changes to the southbound ovsdb-server and that server is
Thanks for the patch, but it's double-spaced and needs to be reposted in
a corrected format.
On Mon, Apr 12, 2021 at 10:32:00AM +, 王亮 Liang Wang wrote:
> From d5c454bc9c68a6b2b40a2301a21f7337d6086d4c Mon Sep 17 00:00:00 2001
>
> From: Wang Liang
>
> Date: Mon, 12 Apr 2021 18:24:21 +0800
>
On Mon, Apr 12, 2021 at 01:36:11PM +0300, Alexey Roytman wrote:
> From: Alexey Roytman
>
> For big scale deployments, when number of logical flows can be 2M+,
> sometimes users just need to know the total number of logical flows
> and numbers of logical flows per table/per datapath.
This isn't
Hi Ali and Han, you guys wrote the docker stuff in OVS, and Numan, you
reviewed it. Do you think you could comment on these proposed changes
from Aidan/hunchback? I am not competent in Docker myself.
On Fri, Apr 02, 2021 at 07:27:00PM +, hunchback wrote:
> add a set of docker targets which
On Fri, Apr 02, 2021 at 07:34:43PM +, hunchback wrote:
> the current documentation assumes everything is run from within root
> what is not correct in the general case, the revised install
> documentation makes use of `sudo` for specific commands which require to
> be run under root
On Fri, Apr 02, 2021 at 07:08:19PM +, hunchback wrote:
> When running repeated builds using `make build` you get prompts in cases
> the `mv` command is about to overwrite a file which is write-protect.
> This patch forced the `mv` w/o prompting for approval.
>
> Signed-off-by: hunchback
I
On Wed, Apr 07, 2021 at 10:27:25PM +0100, Mark Gray wrote:
> On 07/04/2021 20:43, Aidan Shribman wrote:
> > From: hunchback
> >
> > add documentation of install sequence sudo to support running install
> > sequence as non-root user
> >
> > Signed-off-by: Aidan Shribman
> > ---
> >
> >>> +if (id != NULL) {
> >>
> >>It's safe to call free(NULL), so , please, don't check.
> > [Rick] Does it mean that we override the original free() method, so that it
> > won't crash when we call free(NULL)? If so, that is good and I don't need
> > to check here.
>
> It's part of a C
On Tue, Apr 06, 2021 at 08:11:35AM +, hunchback wrote:
> add documentation of install sequence sudo to support running install
> sequence as non-root user
>
> Signed-off-by: hunchback
Thank you, this looks almost perfect. I have only one comment: would
you please sign off with your name,
On Thu, Apr 01, 2021 at 09:54:10AM -0700, Ben Pfaff wrote:
> On Thu, Apr 01, 2021 at 06:33:21PM +0200, Dumitru Ceara wrote:
> > On 4/1/21 5:48 PM, Ben Pfaff wrote:
> > > On Wed, Mar 24, 2021 at 04:03:07PM +0100, Dumitru Ceara wrote:
> > >> Hi Ben,
> > >>
&
instantly on a big database, whereas before it could take
arbitrarily long.
Signed-off-by: Ben Pfaff
---
This goes on top of the previous series.
utilities/ovn-dbctl.c |2 +-
utilities/ovn-nbctl.c | 1028 +++--
2 files changed, 900 insertions(+), 130 deletion
On Wed, Apr 07, 2021 at 01:19:09PM +0200, Ilya Maximets wrote:
> On 3/26/21 9:18 PM, Ben Pfaff wrote:
> > On Fri, Mar 26, 2021 at 07:52:18PM +0100, Ilya Maximets wrote:
> >> On 3/26/21 7:30 PM, Ben Pfaff wrote:
> >>> From: James Page
> >>>
> >>
On Wed, Apr 07, 2021 at 01:12:11PM +0200, Ilya Maximets wrote:
> On 3/26/21 7:30 PM, Ben Pfaff wrote:
> > The thread-local data allocators can't increment coverage counters
> > because this can cause reentrancy. Until now, this code has used
> > explicit calls to malloc()
On Wed, Apr 07, 2021 at 12:58:21PM +0200, Ilya Maximets wrote:
> On 3/26/21 7:30 PM, Ben Pfaff wrote:
> > lldpd_alloc_hardware() always returns nonnull.
> >
> > At the same time, there's no reason that lldpd_alloc_hardware() doesn't
> > take a const char *, so chang
On Wed, Apr 07, 2021 at 02:21:19PM +0200, Ilya Maximets wrote:
> Also, it seems like none of these commands documented anywhere.
> We will need to fix that someday.
I'd recommend requiring documentation for anything being added, so the
situation doesn't get worse.
We usually document this kind
On Mon, Apr 05, 2021 at 10:50:07PM +, hunchback wrote:
> - $ mkdir -p /usr/local/etc/ovn
> + $ sudo mkdir -m 777 -p /usr/local/etc/ovn
Thanks for the patch!
Here and in another case later, it seems like a really bad idea to make
directories world-writable.
On Mon, Apr 05, 2021 at 11:36:42AM -0400, Mark Michelson wrote:
> On 4/5/21 11:18 AM, Ben Pfaff wrote:
> > > > > > - OVSDIR=$srcdir/ovs
> > > > > > + OVSDIR=$(cd $srcdir/ovs; pwd)
> >
> > This transformation less safe than it loo
> > > > - OVSDIR=$srcdir/ovs
> > > > + OVSDIR=$(cd $srcdir/ovs; pwd)
This transformation less safe than it looks at first. Since = doesn't
do word splitting,
OVSDIR=$srcdir/ovs
doesn't need quotes on the right. The "cd" command (and all other
commands) does do word splitting, so it
On Fri, Apr 02, 2021 at 07:11:22PM +, hunchback wrote:
> the current documentation assumes everything is run from within root
> what is not correct in the general case, the revised install
> documentation makes use of `sudo` for specific commands which require to
> be run under root
: Dumitru Ceara
Seems reasonable. I didn't review the specifics.
Acked-by: Ben Pfaff
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
above).
- After removing the field, we observe that changes to intermediate
relations no longer grow with the topology, and transaction timing
increases much more slowly:
https://gist.github.com/ryzhyk/d02784b9088d82f8549ea1b2ebdf095e
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben
From: Leonid Ryzhyk
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lswitch.dl | 1 -
1 file changed, 1 deletion(-)
diff --git a/northd/lswitch.dl b/northd/lswitch.dl
index aff265bbf82c..973faec5073f 100644
--- a/northd/lswitch.dl
+++ b/northd/lswitch.dl
@@ -23,7 +23,6
-table` flag to eliminate all intermediate tables in
`helpers.dl`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 36
northd/lrouter.dl| 12 ++--
northd/lswitch.dl| 26 +-
northd/ovn-nb
From: Leonid Ryzhyk
Use the `--intern-table` switch to intern `Logical_Switch_Port` records,
so that they can be copied and compared efficiently by pointer.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 2 +-
northd/lrouter.dl| 2 +-
northd/lswitch.dl
This is originally from Numan Siddique. I have adapted it a bit
to run faster by using the ovn-nbctl and ovn-sbctl daemons and
combining multiple calls into just one.
I'm uncertain whether to actually commit this; I need a sign-off
from Numan to do so.
Signed-off-by: Ben Pfaff
CC: Numan
From: Leonid Ryzhyk
Change the type of record in the `RouterPort` table from
`Ref` to `Intern`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lrouter.dl| 37 +
northd/lswitch.dl| 4 ++--
northd/ovn_northd.dl | 14
From: Leonid Ryzhyk
Use the `--intern-table` switch to intern `Logical_Router_Port` records,
so that they can be copied and compared efficiently by pointer.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 2 +-
northd/lrouter.dl| 50
From: Leonid Ryzhyk
Change the type of record in the `SwitchPort` table from
`Ref` to `Intern`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ipam.dl | 22 +++
northd/lswitch.dl| 64 +++-
northd/ovn_northd.dl
eeds `&`, we bind
`v` to the value of the smart pointer and not the type that it wraps.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ipam.dl | 14 +++---
northd/multicast.dl | 4 +-
northd/ovn_northd.dl | 104 +--
3 file
From: Leonid Ryzhyk
Change the type of record in the `Switch` table from `Ref` to
`Intern`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lswitch.dl | 37 +
northd/multicast.dl | 10 +-
2 files changed, 26 insertions(+), 21
From: Leonid Ryzhyk
Change all remaining occurrences of `Ref` to `Intern` throughout
the DDlog code base.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 36 +-
northd/lrouter.dl| 22
northd/lswitch.dl| 61
,
var port_names = port_name.group_by((_uuid, sb_name)).to_set().
```
Again, benchmarking is likely to reveal more instances of this. A
proper fix will require a change to the DDlog compiler.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ovn_northd.dl | 19 +--
of the `Switch` object, which can cause
a bunch of rules to update their outputs.
As a workaround, we no longer store the entire `Logical_Switch` object
in the `Switch` table, and instead only copy its relevant fields.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ipam.dl
in the `Router` table, and instead only copy its relevant fields.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lrouter.dl| 47 --
northd/ovn_northd.dl | 374 +--
2 files changed, 220 insertions(+), 201 deletions(-)
diff --git a/northd
Also rewrite the manpage and convert it to XML for consistency with
ovn-nbctl, and add tests.
Signed-off-by: Ben Pfaff
---
NEWS | 4 +-
manpages.mk | 17 -
tests/ovn-sbctl.at| 76 +++--
utilities/automake.mk | 7 +-
utilities/ovn-dbctl.c
do not trigger any other rules.
[1]
https://github.com/vmware/differential-datalog/blob/master/doc/tutorial/tutorial.md#reference-type-ref
[2]
https://github.com/vmware/differential-datalog/blob/master/doc/tutorial/tutorial.md#interned-values-intern-istring
Signed-off-by: Leonid Ryzhyk
In an upcoming commit, this will allow adding daemon mode to ovn-sbctl
without having a lot of duplicated code.
Signed-off-by: Ben Pfaff
---
utilities/automake.mk |5 +-
utilities/ovn-dbctl.c | 1214
utilities/ovn-dbctl.h | 60 ++
utilities/ovn-nbctl.c
be more robust than the
previous version regardless of whether it fixes the exact problem
that Dumitru observed (which I did not try to reproduce).
Reported-by: Dumitru Ceara
Signed-off-by: Ben Pfaff
---
northd/ovn_northd.dl | 63 ++--
1 file changed, 37
over a
`Group` in a for-loop, the iterator returns `(value, weight)` tuples.
This also adapts ovn-northd-ddlog.c to a slightly updated C API.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
NEWS | 2 +-
northd/lrouter.dl | 2 +-
northd/ovn-northd-ddlog.c
Signed-off-by: Ben Pfaff
---
tests/ovn.at | 67 ++--
1 file changed, 49 insertions(+), 18 deletions(-)
diff --git a/tests/ovn.at b/tests/ovn.at
index 391a8bcd9323..7b6789125ffc 100644
--- a/tests/ovn.at
+++ b/tests/ovn.at
@@ -9467,9 +9467,10
This isn't notable, since this commit frees it just before exiting, but
it cleans up the Address Sanitizer report.
Signed-off-by: Ben Pfaff
---
utilities/ovn-dbctl.c | 4
1 file changed, 4 insertions(+)
diff --git a/utilities/ovn-dbctl.c b/utilities/ovn-dbctl.c
index 28ebc6267066
in a few places.
Signed-off-by: Ben Pfaff
---
utilities/ovn-nbctl.8.xml | 670 ++
1 file changed, 392 insertions(+), 278 deletions(-)
diff --git a/utilities/ovn-nbctl.8.xml b/utilities/ovn-nbctl.8.xml
index 03d47dba5b72..39f9381fdaae 100644
--- a/utilities/ovn
Mark says that it was heavily used by ovn-kubernetes for years without
any issues.
Signed-off-by: Ben Pfaff
Suggested-by: Mark Michelson
---
NEWS | 1 +
utilities/ovn-nbctl.8.xml | 4
2 files changed, 1 insertion(+), 4 deletions(-)
diff --git a/NEWS b/NEWS
index
They do exactly the same thing but ovn-appctl has more ovn in its name.
Signed-off-by: Ben Pfaff
Suggested-by: Mark Michelson
---
utilities/ovn-nbctl.8.xml | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/utilities/ovn-nbctl.8.xml b/utilities/ovn-nbctl.8.xml
index
ory leak in client mode."
and "ovn-northd-ddlog: Fix two memory leaks." fix memory leaks
reported by Numan and found by Address Sanitizer.
- Fix bug introduced into ovsdb2ddlog2c in "ovn-northd-ddlog: Intern
selected input relations."
Ben Pfaff (11):
ovn-n
This isn't notable, since this commit frees it just before exiting, but
it cleans up the Address Sanitizer report.
Signed-off-by: Ben Pfaff
---
utilities/ovn-nbctl.c | 4
1 file changed, 4 insertions(+)
diff --git a/utilities/ovn-nbctl.c b/utilities/ovn-nbctl.c
index 18405835699d
I get a clean report from Address Sanitizer now.
Signed-off-by: Ben Pfaff
---
northd/ovn-northd-ddlog.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/northd/ovn-northd-ddlog.c b/northd/ovn-northd-ddlog.c
index b7d2c8a5ef8d..ca1ab325448c 100644
--- a/northd/ovn-northd-ddlog.c
+++ b
On Thu, Apr 01, 2021 at 06:33:21PM +0200, Dumitru Ceara wrote:
> On 4/1/21 5:48 PM, Ben Pfaff wrote:
> > On Wed, Mar 24, 2021 at 04:03:07PM +0100, Dumitru Ceara wrote:
> >> Hi Ben,
> >>
> >> We discussed a bit about this during one of the recent IRC OVN me
On Wed, Mar 24, 2021 at 04:03:07PM +0100, Dumitru Ceara wrote:
> Hi Ben,
>
> We discussed a bit about this during one of the recent IRC OVN meetings,
> but I didn't get around to properly reporting this until now.
>
> I've tried running ovn-northd-ddlog against some large OVN NB/DB
> databases
I found the leaks and the fixes will be in v2 of the series.
On Thu, Mar 25, 2021 at 01:14:37PM -0700, Ben Pfaff wrote:
> OK. I guess I'll have to try it myself.
>
> On Thu, Mar 25, 2021 at 12:59:00PM +0530, Numan Siddique wrote:
> > Hi Ben,
> >
> > With your memory
On Wed, Mar 31, 2021 at 09:35:36AM +0200, Dumitru Ceara wrote:
> On 3/30/21 11:50 PM, Ben Pfaff wrote:
> > Without this change, ovsdb2ddlog2c exited successfully if it ran into an
> > option that it was supposed to understand but didn't implement (which
> > is a bug). This
On Wed, Mar 31, 2021 at 07:18:03PM +0200, Paolo Valerio wrote:
> this patch introduces for the userspace datapath the handling
> of rules like the following:
>
> ct(commit,nat(src=0.0.0.0),...)
>
> Kernel datapath already handle this case that is particularly
> handy in scenarios like the
On Mon, Mar 29, 2021 at 12:50:56AM +0300, Alexey Roytman wrote:
> From: Alexey Roytman
>
> Signed-off-by: Alexey Roytman
Thanks. I squashed these two patches together (features and their tests
should be in a single patch) and applied it to master.
On Tue, Mar 30, 2021 at 03:29:54PM -0700, Jean Tourrilhes wrote:
> Have fun...
>From that line you can be sure that this is a genuine Jean Tourrilhes
email.
Thanks, Jean!
___
dev mailing list
d...@openvswitch.org
On Mon, Mar 29, 2021 at 07:42:53PM -0400, Mark Michelson wrote:
> Thank you, thank you, thank you for converting the ovn-sb documentation to
> XML.
>
> On 3/25/21 7:26 PM, Ben Pfaff wrote:
> > + Daemon mode is internally implemented using the same mechanism used
> >
On Mon, Mar 29, 2021 at 07:42:47PM -0400, Mark Michelson wrote:
> On 3/25/21 7:26 PM, Ben Pfaff wrote:
> > This rearranges the manpage into a more logical order, documents some
> > options that weren't documented, adds some sections such as
> > Environment and Exit Status
Without this change, ovsdb2ddlog2c exited successfully if it ran into an
option that it was supposed to understand but didn't implement (which
is a bug). This commit makes it raise an exception instead.
Signed-off-by: Ben Pfaff
Reported-by: Dumitru Ceara
---
northd/ovsdb2ddlog2c | 2 +-
1
On Tue, Mar 30, 2021 at 05:52:05PM +0200, Dumitru Ceara wrote:
> On 3/27/21 1:31 AM, Ben Pfaff wrote:
> > From: Leonid Ryzhyk
> >
> > DDlog 0.38.0 adds the `--intern-table` CLI flag to the `ovsdb2ddlog`
> > compiler to declare input tables coming from OVSDB as `Inter
On Tue, Mar 30, 2021 at 11:16:48PM +0200, Ilya Maximets wrote:
> On 3/3/21 3:46 PM, xiangxia.m@gmail.com wrote:
> > From: Tonghao Zhang
> >
> > If user don't set the meter "burst_size", when creating them. OvS
> > will set "brust_size" to "rate", and there will be a double "rate"
> > as
On Tue, Mar 30, 2021 at 02:05:40PM +0100, Aidan Shribman wrote:
> add a set of docker targets which ease the building and testing of
> docker images both on local registry (localhost:5000) as well as on the
> public default registry (docker.io).
>
> Signed-off-by: Aidan Shribman
Thanks for the
On Tue, Mar 30, 2021 at 11:55:11AM +0100, Aidan Shribman wrote:
> the current documentation assumes everything is run from within root
> what is not correct in the general case, the revised install
> documentation makes use of `sudo` for specific commands which require to
> be run under root
On Tue, Mar 30, 2021 at 11:43:34AM +0100, Aidan Shribman wrote:
> When running repeated builds using `make build` you get prompts in cases
> the `mv` command is about to overwrite a file which is write-protect. This
> patch forced the `mv` w/o prompting for approval.
>
> Signed-off-by: Aidan
On Tue, Mar 30, 2021 at 11:31:35AM -0500, Ansis wrote:
> On Mon, Mar 29, 2021 at 6:21 PM Ben Pfaff wrote:
> >
> > On Mon, Mar 29, 2021 at 05:26:34PM -0500, Ansis Atteka wrote:
> > > Under high load I observed that Netlink buffer constantly
> > > fills up for
On Mon, Mar 29, 2021 at 05:26:34PM -0500, Ansis Atteka wrote:
> Under high load I observed that Netlink buffer constantly
> fills up for daemons querying Conntrack Table that has a
> lot of entries in it:
>
> netlink_notifier|WARN|netlink receive buffer overflowed
>
> This patch mitigates the
above).
- After removing the field, we observe that changes to intermediate
relations no longer grow with the topology, and transaction timing
increases much more slowly:
https://gist.github.com/ryzhyk/d02784b9088d82f8549ea1b2ebdf095e
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben
This is originally from Numan Siddique. I have adapted it a bit
to run faster by using the ovn-nbctl and ovn-sbctl daemons and
combining multiple calls into just one.
I'm uncertain whether to actually commit this; I need a sign-off
from Numan to do so.
Signed-off-by: Ben Pfaff
CC: Numan
-table` flag to eliminate all intermediate tables in
`helpers.dl`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 36
northd/lrouter.dl| 12 ++--
northd/lswitch.dl| 26 +-
northd/ovn-nb
From: Leonid Ryzhyk
Use the `--intern-table` switch to intern `Logical_Router_Port` records,
so that they can be copied and compared efficiently by pointer.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 2 +-
northd/lrouter.dl| 50
From: Leonid Ryzhyk
Use the `--intern-table` switch to intern `Logical_Switch_Port` records,
so that they can be copied and compared efficiently by pointer.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 2 +-
northd/lrouter.dl| 2 +-
northd/lswitch.dl
eeds `&`, we bind
`v` to the value of the smart pointer and not the type that it wraps.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ipam.dl | 14 +++---
northd/multicast.dl | 4 +-
northd/ovn_northd.dl | 104 +--
3 file
From: Leonid Ryzhyk
Change all remaining occurrences of `Ref` to `Intern` throughout
the DDlog code base.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/helpers.dl| 36 +-
northd/lrouter.dl| 22
northd/lswitch.dl| 61
of the `Switch` object, which can cause
a bunch of rules to update their outputs.
As a workaround, we no longer store the entire `Logical_Switch` object
in the `Switch` table, and instead only copy its relevant fields.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ipam.dl
From: Leonid Ryzhyk
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lswitch.dl | 1 -
1 file changed, 1 deletion(-)
diff --git a/northd/lswitch.dl b/northd/lswitch.dl
index aff265bbf82c..973faec5073f 100644
--- a/northd/lswitch.dl
+++ b/northd/lswitch.dl
@@ -23,7 +23,6
From: Leonid Ryzhyk
Change the type of record in the `RouterPort` table from
`Ref` to `Intern`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lrouter.dl| 37 +
northd/lswitch.dl| 4 ++--
northd/ovn_northd.dl | 14
From: Leonid Ryzhyk
Change the type of record in the `Switch` table from `Ref` to
`Intern`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lswitch.dl | 37 +
northd/multicast.dl | 10 +-
2 files changed, 26 insertions(+), 21
From: Leonid Ryzhyk
Change the type of record in the `SwitchPort` table from
`Ref` to `Intern`.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ipam.dl | 22 +++
northd/lswitch.dl| 64 +++-
northd/ovn_northd.dl
,
var port_names = port_name.group_by((_uuid, sb_name)).to_set().
```
Again, benchmarking is likely to reveal more instances of this. A
proper fix will require a change to the DDlog compiler.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/ovn_northd.dl | 19 +--
in the `Router` table, and instead only copy its relevant fields.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
northd/lrouter.dl| 47 --
northd/ovn_northd.dl | 374 +--
2 files changed, 220 insertions(+), 201 deletions(-)
diff --git a/northd
do not trigger any other rules.
[1]
https://github.com/vmware/differential-datalog/blob/master/doc/tutorial/tutorial.md#reference-type-ref
[2]
https://github.com/vmware/differential-datalog/blob/master/doc/tutorial/tutorial.md#interned-values-intern-istring
Signed-off-by: Leonid Ryzhyk
be more robust than the
previous version regardless of whether it fixes the exact problem
that Dumitru observed (which I did not try to reproduce).
Reported-by: Dumitru Ceara
Signed-off-by: Ben Pfaff
---
northd/ovn_northd.dl | 63 ++--
1 file changed, 37
over a
`Group` in a for-loop, the iterator returns `(value, weight)` tuples.
This also adapts ovn-northd-ddlog.c to a slightly updated C API.
Signed-off-by: Leonid Ryzhyk
Signed-off-by: Ben Pfaff
---
NEWS | 2 +-
northd/lrouter.dl | 2 +-
northd/ovn-northd-ddlog.c
In an upcoming commit, this will allow adding daemon mode to ovn-sbctl
without having a lot of duplicated code.
Signed-off-by: Ben Pfaff
---
utilities/automake.mk |5 +-
utilities/ovn-dbctl.c | 1214
utilities/ovn-dbctl.h | 60 ++
utilities/ovn-nbctl.c
Also rewrite the manpage and convert it to XML for consistency with
ovn-nbctl, and add tests.
Signed-off-by: Ben Pfaff
---
NEWS | 2 +
manpages.mk | 17 -
tests/ovn-sbctl.at| 76 +++--
utilities/automake.mk | 7 +-
utilities/ovn-dbctl.c
in a few places.
Signed-off-by: Ben Pfaff
---
utilities/ovn-nbctl.8.xml | 670 ++
1 file changed, 392 insertions(+), 278 deletions(-)
diff --git a/utilities/ovn-nbctl.8.xml b/utilities/ovn-nbctl.8.xml
index 03d47dba5b72..39f9381fdaae 100644
--- a/utilities/ovn
Signed-off-by: Ben Pfaff
---
tests/ovn.at | 67 ++--
1 file changed, 49 insertions(+), 18 deletions(-)
diff --git a/tests/ovn.at b/tests/ovn.at
index 391a8bcd9323..7b6789125ffc 100644
--- a/tests/ovn.at
+++ b/tests/ovn.at
@@ -9467,9 +9467,10
... 1.804s)
ddlog after optimization patches: 35.2s (0.129s ... 0.147s)
I previously posted the first three patches. This series includes
those patches with the minor fixes that I posted later folded in.
Ben Pfaff (6):
ovn-nbctl: Improve manpage.
ovn-nbctl: Refactor into infrastructure
This is the first three patches of the series that I'm about to post
with further optimizations.
On Thu, Mar 25, 2021 at 04:26:05PM -0700, Ben Pfaff wrote:
> A daemon mode for ovn-sbctl can make benchmarks much faster. For me,
> it made the benchmark provided by Numan drop from several m
On Thu, Mar 25, 2021 at 04:26:08PM -0700, Ben Pfaff wrote:
> Also rewrite the manpage and convert it to XML for consistency with
> ovn-nbctl, and add tests.
>
> Signed-off-by: Ben Pfaff
Here's a fix that needs to be folded in:
diff --git a/utilities/ovn-sbctl.c b/utilities/ovn-s
301 - 400 of 24009 matches
Mail list logo