On Mon, 19 Apr 2021 11:58:22 +0530 madhukar mythri <madhukar.myt...@gmail.com> wrote:
> Hi Stephen, > > Thanks for quick response. > Yes this error is from a TAP driver from DPDK, but why this error/issue > occurs only when configured with -2queues only ? > Whereas with single or 4 queues it works Rx/Tx traffic works well. > > And also, this issue occurs only on single port-0, whereas on other ports > this error/issue is not seen. > > > On Sun, Apr 18, 2021 at 9:08 PM Stephen Hemminger < > step...@networkplumber.org> wrote: > > > On Sun, 18 Apr 2021 17:51:09 +0530 > > madhukar mythri <madhukar.myt...@gmail.com> wrote: > > > > > Hi, > > > > > > When we tried to launch a DPDK app on Azure VM with 2-queues, > > > seeing following errors and thus we are not able to receive any traffic > > > on these NIC ports(MLX5). > > > On Azure VM, using "net_failsafe" PMD. > > > > > > ======================= > > > PORT 0 Max supports 16 rx queues and 16 tx queues (driver_name = > > > net_failsafe, driver_type = 16) > > > PORT 0 is polling for link-change, interrupts disabled > > > *rss_add_actions(): Failed to load BPF section l3_l4 (11): Resource > > > temporarily unavailable* > > > net_failsafe: Failed to create flow on sub_device 1 > > > add_flow(): create() fails for port 0; Reason: action not supported > > > =================== > > > > > > This issue is seen with 2-queues only, whereas with 4-queues we are not > > > seeing this issue. > > > > > > Does anybody faced similar issue and if any solution/work-around please > > let > > > me know. > > > > > > DPDK version: 20.11 > > > With RHEL: 7.9 version. > > > > > > Thanks, > > > Madhukar. > > > > > > If you are using failsafe, make sure that you have the Mellanox drivers > > enabled > > in your DPDK build, also TAP device. You need to build with rdma-core > > version 23 or > > later, and you need to have the kernel modules for flower, multiq, and TAP. > > > > Turn on full DPDK logging and check the kernel log (dmesg). > > No idea, look at the kernel log. Might be related to RSS.
