On 2/10/25 7:31 AM, Ales Musil wrote: > On Mon, Feb 10, 2025 at 7:17 AM Ales Musil <[email protected]> wrote: > >> >> >> On Fri, Feb 7, 2025 at 9:31 PM Dumitru Ceara <[email protected]> wrote: >> >>> On 2/7/25 5:31 PM, Numan Siddique wrote: >>>> On Thu, Feb 6, 2025 at 9:19 AM Felix Huettner via dev >>>> <[email protected]> wrote: >>>>> >>>>> This brings in needed changes so that the ovs submodule exports route >>>>> table related functionality. >>>>> >>>>> As the ovs submodule has bumped its DPDK version we need to do so as >>>>> well. >>>>> >>>>> Acked-by: Dumitru Ceara <[email protected]> >>>>> Signed-off-by: Felix Huettner <[email protected]> >>>>> --- >>>>> ovs | 2 +- >>>>> utilities/containers/prepare.sh | 2 +- >>>>> 2 files changed, 2 insertions(+), 2 deletions(-) >>>>> >>>>> diff --git a/ovs b/ovs >>>>> index 7b1ce8e2a..57c3cc37c 160000 >>>>> --- a/ovs >>>>> +++ b/ovs >>>>> @@ -1 +1 @@ >>>>> -Subproject commit 7b1ce8e2a08454839c52b8cc02fdde5c78e7c40e >>>>> +Subproject commit 57c3cc37c8a3c08426a29486df7a03da84907fa2 >>>> >>>> Lets update to this commit 63785d16db11ff9f5d5f02595cb5bbe400e4de9b >>>> (or the tip of branch-3.5). >>>> 63785d16db11ff9f5d5f02595cb5bbe400e4de9b is just one commit above >>>> 57c3cc37c8a3c08426a29486df7a03da84907fa2 >>>> and we need it so that we can remove a hack in >>> .ci/ovn-kubernetes/Dockerfile >>>> >>>> If there will be v8, please update it. Otherwise I guess Dumitru or >>>> others who merge can update it. >>>> >>> >>> Thanks for pointing it out, Numan! I went ahead and applied this patch >>> to main. I made the suggested change and moved the submodule to the >>> current tip of branch-3.5: >>> >>> >>> https://github.com/ovn-org/ovn/commit/86dc9f1574ba026fb3d4189f60bd35f6cf8b1c6a >>> >>> Hope it's fine with you too, Felix. >>> >>> Regards, >>> Dumitru >>> >>>> Thanks >>>> Numan >>>> >>>> >>>>> diff --git a/utilities/containers/prepare.sh >>> b/utilities/containers/prepare.sh >>>>> index 49bb2ac91..6922d8c6b 100755 >>>>> --- a/utilities/containers/prepare.sh >>>>> +++ b/utilities/containers/prepare.sh >>>>> @@ -1,7 +1,7 @@ >>>>> #!/bin/bash -xe >>>>> >>>>> DPDK_GIT=https://dpdk.org/git/dpdk >>>>> -DPDK_VER=23.11 >>>>> +DPDK_VER=24.11 >>>>> >>>>> function compile_sparse() >>>>> { >>>>> -- >>>>> 2.47.1 >>>>> >>>>> >>>>> _______________________________________________ >>>>> dev mailing list >>>>> [email protected] >>>>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev >>>> _______________________________________________ >>>> dev mailing list >>>> [email protected] >>>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev >>> >>> _______________________________________________ >>> dev mailing list >>> [email protected] >>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev >> >> >> Hi, >> >> it seems like the DPDK bump didn't help, our CI is failing with the >> following: >> >> lib/netdev-dpdk.c:6440:19: error: 'rte_vhost_driver_set_max_queue_num' is >> deprecated: Symbol is not yet part of stable ABI >> [-Werror,-Wdeprecated-declarations] >> 6440 | err = rte_vhost_driver_set_max_queue_num(dev->vhost_id, >> max_qp); >> | ^ >> /workspace/dpdk-dir/include/rte_vhost.h:624:1: note: >> 'rte_vhost_driver_set_max_queue_num' has been explicitly marked deprecated >> here >> 624 | __rte_experimental >> | ^ >> /workspace/dpdk-dir/include/rte_compat.h:19:16: note: expanded from macro >> '__rte_experimental' >> 19 | __attribute__((deprecated("Symbol is not yet part of stable ABI"), \ >> | ^ >> 1 error generated. >> make[1]: *** [Makefile:4771: lib/netdev-dpdk.lo] Error 1 >> make[1]: *** Waiting for unfinished jobs.... >> >> >> Seems like the same error that Numan reported on the Fedora bump. >> Hopefully we will be able to resolve it quickly. >> >> Regards, >> Ales >> > > It's due to the old container image which doesn't have the new DPDK version > yet, which is slightly problematic because the container job is failing for > a while. For now it affects only periodic jobs and jobs after merge. >
Hi Ales, Yes, it's the same problem we hit last week with random crashes in the ARM64 container jobs executed on GitHub runners. I can't reproduce any of this locally. What if we "manually" run an x86 job to at least have some CI running with dpdk (on x86) until the ARM64 on x86 GHA runners get resolved. I can push this commit to a temporary branch on the ovn-org repo: https://github.com/dceara/ovn/commit/72a77f5805e03f6d77833a7447e97dffc3d5e614 Then trigger a manual run of the containers workflow on that branch. That will push the x86 image to ghcr.io, after that I can just remove the temporary branch. I could also do the push manually but we already have most of it automated so I'd do that. What do you think? Thanks, Dumitru _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
