Re: [ovs-dev] [PATCH ovn v3 2/2] ci: Pin Python, Fedora and Ubuntu runner versions.

On 11/16/23 19:04, Dumitru Ceara wrote: >>> diff --git a/utilities/containers/fedora/Dockerfile >>> b/utilities/containers/fedora/Dockerfile >>> index 4058d7f5be..066bb0b957 100755 >>> --- a/utilities/containers/fedora/Dockerfile >>> +++ b/utilities/containers/fedora/Dockerfile >>> @@ -1,4 +1,4 @@

Re: [ovs-dev] [PATCH ovn v3 0/2] Stabilize CI by pinning container, runner and Python versions.

On 11/16/23 21:48, Mark Michelson wrote: > On 11/16/23 12:35, Ales Musil wrote: >> >> >> On Thu, Nov 16, 2023 at 6:26 PM Dumitru Ceara > > wrote: >> >>     All of the above were changed to track the latest available releases. >>     Initially that seemed like a good idea

Re: [ovs-dev] [PATCH ovn v3 0/3] handle pb->chassis and pb->up from if-status module

On 10/30/23 10:10, Xavier Simonart wrote: > Xavier Simonart (3): > binding: slight refactor if no local binding in consider_iface_release > binding: handle pb->chassis and pb->up from if-status module > Fix flows not removed in ha migration > > controller/binding.c| 221

Re: [ovs-dev] [PATCH ovn 0/5] More unit tests fixes

On 11/3/23 14:42, Xavier Simonart wrote: > Xavier Simonart (5): > tests: fixed "interconnection - static multicast" and "- IGMP/MLD > multicast" > tests: fixed system test "LR with SNAT fragmentation needed for > external server". > tests: fixed "ovn-nbctl - daemon retry connection"

Re: [ovs-dev] [PATCH ovn] treewide: Avoid empty initializer

On 11/2/23 15:40, Ihar Hrachyshka wrote: > +1, thanks for the cleanup. > > Reviewed-By: Ihar Hrachyshka > Thanks, Ales and Ihar! Applied to main. Regards, Dumitru > On Thu, Nov 2, 2023 at 8:50 AM Ales Musil wrote: > >> The empty initializer (" = {}") is allowed only from >> C23 and

Re: [ovs-dev] [PATCH ovn] tests: fixed another set of flaky ovn-ic tests

On 11/3/23 10:26, Ales Musil wrote: > On Mon, Oct 23, 2023 at 12:05 PM Xavier Simonart > wrote: > >> Some ic tests were failing as not waiting, or not waiting properly >> for some information to be available in the az: >> - Adding a port to ts1 before ts1 being created in that az. >> - Setting

Re: [ovs-dev] [PATCH ovn v2] controller: have I+P assigning ct_zones for l3gateway ports

On 10/30/23 09:09, Xavier Simonart wrote: > When l3gateway ports get added, ct_zones were assigned during > (ct_zones) recomputes, but not by I+P. > Before this patch, test "Migration of CT zone from UUID to name" > was randomly failing, as ct_zone was not assigned by I+P but > a ct_zone recompute

Re: [ovs-dev] [PATCH ovn] ovn-ctl man: Add election timer config to manpage

On 10/31/23 14:49, martin.kal...@canonical.com wrote: > > Raft election timers for NB and SB DB clusters are configurable via > ovn-ctl but the options are not mention on the man page. > > Signed-off-by: Martin Kalcok > --- >  utilities/ovn-ctl.8.xml | 2 ++ >  1 file changed, 2 insertions(+) >

Re: [ovs-dev] [PATCH ovn v2 0/5] Fixed another set of flaky Unit Tests

On 11/3/23 10:23, Ales Musil wrote: > On Mon, Oct 23, 2023 at 11:23 AM Xavier Simonart > wrote: > >> Xavier Simonart (5): >> tests: fixed multiple tests not properly waiting for packets to be >> received >> tests: do not start backup-northd by default >> tests: fixed "LSP incremental

Re: [ovs-dev] [PATCH ovn] fake-multinode-ci: Use python version '3.11' instead of '3.x'.

On 11/13/23 14:37, Dumitru Ceara wrote: > On 11/9/23 22:47, Dumitru Ceara wrote: >> On 11/7/23 22:33, Mark Michelson wrote: >>> On 11/6/23 23:47, Dumitru Ceara wrote: On 11/2/23 16:30, num...@ovn.org wrote: > From: Numan Siddique > > With the 'ubuntu-latest' image in github CI, 

Re: [ovs-dev] [RFC] netlink-conntrack: optimize flushing ct zone

Bleep bloop. Greetings Felix Huettner, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should start with a capital. WARNING: The subject summary

[ovs-dev] [RFC] netlink-conntrack: optimize flushing ct zone

NOTE: this change makes improvements depending on a change in the kernel currently on the netdev mailing list (see [1]). Previously the kernel did not provide a netlink interface to flush/list only conntrack entries matching a specific zone. With [1] it is now possible to flush and list conntrack

[ovs-dev] [PATCH 4/4] Handle EAGAIN on unix socket connect() failure

While POSIX document EINPROGRESS, Linux connect() system call returns EAGAIN for nonblocking Unix sockets instead. This patch handles this case by repeating connect() attempts. Signed-off-by: Ihar Hrachyshka --- lib/socket-util-unix.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff

[ovs-dev] [PATCH 1/4] Use listen backlog = 64 for all connections

Before the patch, the size of the backlog depended on the type of socket (UNIX vs INET) as well as on the language (C vs Python), specifically: - python used backlog size = 10 for all sockets; - C used 10 for unix sockets but 64 for inet sockets. This consolidates the values across the board. It

[ovs-dev] [PATCH 2/4] Don't convert EAGAIN to EPROTO for unix sockets

Unix sockets also have listener backlog. Returning EAGAIN is valid to allow the caller handle the non-blocking temporary failure to connect() gracefully. Signed-off-by: Ihar Hrachyshka --- lib/socket-util-unix.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/lib/socket-util-unix.c

[ovs-dev] [PATCH 3/4] Loop to connect unix socket until success

In nonblocking mode, a POSIX socket may temporarily fail to connect(), which is indicated by returning EINPROGRESS. This patch will loop waiting for connect() to succeed, or until EINPROGRESS happens. An alternative - and probably a better - path to deal with this error would be to extend unix

[ovs-dev] [PATCH 0/4] Improve unixctl AF_UNIX backlog handling

This series attempts to improve behavior of unixctl clients and servers when AF_UNIX sockets are used. - python servers now use a larger - 64 - backlog size to reflect the size of AF_INET unixctl C servers. - EAGAIN / EINPROGRESS from socket() are handled by looping until success - conversion

[ovs-dev] [PATCH v2 0/4] Improve unixctl AF_UNIX backlog handling

This series attempts to improve behavior of unixctl clients and servers when AF_UNIX sockets are used: - python servers now use a larger - 64 - backlog size to reflect the size of AF_UNIX unixctl C servers. (Also bumped the size for AF_INET C servers.) - EAGAIN / EINPROGRESS from socket()

[ovs-dev] [PATCH v2 1/4] Use listen backlog = 64 for all connections

Before the patch, the size of the backlog depended on the type of socket (UNIX vs INET) as well as on the language (C vs Python), specifically: - python used backlog size = 10 for all sockets; - C used 64 for UNIX sockets but 10 for INET sockets. This consolidates the values across the board. It

[ovs-dev] [PATCH v2 2/4] Don't convert EAGAIN to EPROTO for unix sockets

Unix sockets also have listener backlog. Returning EAGAIN is valid to allow the caller handle the non-blocking temporary failure to connect() gracefully. Signed-off-by: Ihar Hrachyshka --- lib/socket-util-unix.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/lib/socket-util-unix.c

[ovs-dev] [PATCH v2 3/4] Loop to connect unix socket until success

In nonblocking mode, a POSIX socket may temporarily fail to connect(), which is indicated by returning EINPROGRESS. This patch will loop waiting for connect() to succeed, or until EINPROGRESS happens. An alternative - and probably a better - path to deal with this error would be to extend unix

[ovs-dev] [PATCH v2 4/4] Handle EAGAIN on unix socket connect() failure

While POSIX document EINPROGRESS, Linux connect() system call returns EAGAIN for nonblocking Unix sockets instead. This patch handles this case by repeating connect() attempts. Signed-off-by: Ihar Hrachyshka --- lib/socket-util-unix.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff

Re: [ovs-dev] [PATCH 1/4] Use listen backlog = 64 for all connections

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Use listen backlog = 64 for

Re: [ovs-dev] [PATCH 2/4] Don't convert EAGAIN to EPROTO for unix sockets

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Don't convert EAGAIN to EPROTO

Re: [ovs-dev] [PATCH 3/4] Loop to connect unix socket until success

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Loop to connect unix socket

Re: [ovs-dev] [PATCH 4/4] Handle EAGAIN on unix socket connect() failure

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Handle EAGAIN on unix socket

Re: [ovs-dev] [PATCH ovn v4] controller: Don't artificially limit group and meter IDs to 16bit.

On 10/31/23 18:00, Dumitru Ceara wrote: > OVS actually supports way more. Detect the real number of groups and > meters instead. To avoid preallocating huge bitmaps for the IDs, > switch to id-pool instead (as suggested by Ilya). > > Reported-at: https://issues.redhat.com/browse/FDP-70 >

Re: [ovs-dev] [PATCH v2 1/4] Use listen backlog = 64 for all connections

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Use listen backlog = 64 for

Re: [ovs-dev] [PATCH ovn v2 09/18] northd: Add a new node ls_lbacls.

On Wed, Nov 15, 2023 at 1:55 AM Han Zhou wrote: > > On Thu, Oct 26, 2023 at 11:17 AM wrote: > > > > From: Numan Siddique > > > > This new engine now maintains the load balancer and ACL data of a > > logical switch which was earlier part of northd engine node data. > > The main inputs to this

Re: [ovs-dev] [PATCH ovn v2 05/18] northd: Add a new engine 'lr-nat' to manage lr NAT data.

On Wed, Nov 15, 2023 at 1:27 AM Han Zhou wrote: > > On Thu, Oct 26, 2023 at 11:15 AM wrote: > > > > From: Numan Siddique > > > > This new engine now maintains the NAT related data for each > > logical router which was earlier maintained by the northd > > engine node in the 'struct

Re: [ovs-dev] [PATCH v2 2/4] Don't convert EAGAIN to EPROTO for unix sockets

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Don't convert EAGAIN to EPROTO

Re: [ovs-dev] [PATCH v2 3/4] Loop to connect unix socket until success

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Loop to connect unix socket

Re: [ovs-dev] [PATCH v2 4/4] Handle EAGAIN on unix socket connect() failure

Bleep bloop. Greetings Ihar Hrachyshka, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: WARNING: The subject summary should end with a dot. Subject: Handle EAGAIN on unix socket

[ovs-dev] [PATCH ovn 3/4] fmt_pkt: improve scapy-server logging

The daemon will now log to scapy.log file. Log messages include stats on request processing time as well as any errors that may happen during processing. If you'd like to see even more logs (e.g. for debugging purposes), just pass --verbose to scapy-server. Signed-off-by: Ihar Hrachyshka ---

[ovs-dev] [PATCH ovn 1/4] fmt_pkt: don't subshell when calling ovs-appctl

Signed-off-by: Ihar Hrachyshka --- tests/ovn-macros.at | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ovn-macros.at b/tests/ovn-macros.at index 8dc4ec75c..3191bb6ad 100644 --- a/tests/ovn-macros.at +++ b/tests/ovn-macros.at @@ -876,7 +876,7 @@ fmt_pkt() { if [[ !

[ovs-dev] [PATCH ovn 4/4] fmt_pkt: make sure scapy-server is started once

When running fmt_pkt in parallel, before the patch, there could be several attempts to start scapy-server. Using flock on a test case specific file should guarantee that only one of the subshells will actually be able to start a daemon, which is designed to be a singleton. Now that

[ovs-dev] [PATCH ovn 0/4] misc fmt_pkt improvements

This series, combined with ovs series [1], will allow to execute fmt_pkt in parallel. This will allow to revert back to parallel execution of test_ip in a test conversion patch from Mark [2], and speed up the case while at it. (I will post a separate patch to switch back to background test_ip

[ovs-dev] [PATCH ovn 2/4] fmt_pkt: use -S check to wait for scapy sock file

While either check works, it's better to use a more explicit check. Signed-off-by: Ihar Hrachyshka --- tests/ovn-macros.at | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ovn-macros.at b/tests/ovn-macros.at index 3191bb6ad..0b15bcc80 100644 --- a/tests/ovn-macros.at