From 97b0c5d327b256e0de4c978c832c0397d97adb59 Mon Sep 17 00:00:00 2001 From: Konrad Sztyber Date: Fri, 23 Apr 2021 11:04:23 +0200 Subject: [PATCH] doc/nvmf: add note about the issues with RDMA on E810 NICs Added note to the nvmf documentation about an issue [1] with destroying qpairs on E810 cards in RoCE mode. The note needs to be removed once the issue is resolved. [1] https://github.com/spdk/spdk/issues/1840 Signed-off-by: Konrad Sztyber Change-Id: Ie03a7a662391bd738848f71e582748a334a6ffd5 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7577 Tested-by: SPDK CI Jenkins Community-CI: Broadcom CI Community-CI: Mellanox Build Bot Reviewed-by: Tomasz Zawadzki Reviewed-by: John Kariuki Reviewed-by: Jim Harris --- doc/nvmf.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/doc/nvmf.md b/doc/nvmf.md index 4f4b9e991..065c57303 100644 --- a/doc/nvmf.md +++ b/doc/nvmf.md @@ -106,6 +106,10 @@ using 1GB hugepages or by pre-reserving memory at application startup with `--me option. All pre-reserved memory will be registered as a single region, but won't be returned to the system until the SPDK application is terminated. +Another known issue occurs when using the E810 NICs in RoCE mode. Specifically, the NVMe-oF target +sometimes cannot destroy a qpair, because its posted work requests don't get flushed. It can cause +the NVMe-oF target application unable to terminate cleanly. + ## TCP transport support {#nvmf_tcp_transport} The transport is built into the nvmf_tgt by default, and it does not need any special libraries.