https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-12453 Lustre <p> </p> <p>We have encountered a problem when running RoCEv2 over bonded interfaces.<br/> When bond0 interface is created on top of two slave interfaces beloging to separate HCAs and primary interface fails<br/> after RDMA QPs are created LNET connection is not properly re-established using the backup link.</p> <p>In such case only solution is to reenable/fix primary interface or restart lnet by reloading kernel modules.</p> <p>Problem has been seen on ES7990 as well as in vanilla lustre 2.10.*</p> <p>Normaly when bonding is created on top of two ports belonging to the same HCA - mlx driver is handling link failure by moving QPs. In case described above link failure must be handled in ko2iblnd driver.</p> <p>Log message related to the described bug is logged when problem occurs:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> e0-oss03 kernel: LNetError: 4598:0:(o2iblnd.c:831:kiblnd_create_conn()) cmid HCA(mlx5_0), kib_dev(bond0.881) need failover</pre> </div></div> <p> </p> <p> </p> RDMA over Ethernet: <br/> &nbsp;- Mellanox ConnectX-5 adapters , ES7990 <br/> &nbsp;- o2iblnd(bond0.881) <br/> &nbsp;- bond0.881: mlx5_0 + mlx5_1 (interfaces from separate HCAs) <br/> &nbsp;- bond0 type is active-passive <br/> LU-12453

ko2iblnd: problem handling link failures on bonded interfaces

Bug Blocker Open Unresolved WC Triage Lukasz Flis Wed, 19 Jun 2019 10:26:10 +0000 Thu, 3 Oct 2019 03:13:25 +0000 Lustre 2.10.8 0 2 Related Development Rank 1|i00ig7: Rank (Obsolete) 9223372036854775807 Severity