[LU-12413] Lustre don't able to start if one interface down.

• Type: Bug
• Resolution: Duplicate
• Priority: Blocker
• Affects Version/s: Lustre 2.13.0
• Severity: 3
• Assignee: WC Triage
• Reporter: Alexey Lyashkov

Description

Landing upstream code related to interface scan broke a starting lustre in case one or more interfaces in down.

[ 2969.530212] LNetError: 15492:0:(o2iblnd.c:2892:kiblnd_create_dev()) Can’t query IPoIB interface eth2: it’s down

attached fix works for me.

diff --git a/lnet/klnds/o2iblnd/o2iblnd.c b/lnet/klnds/o2iblnd/o2iblnd.c
index 6da280e..f831a46 100644
--- a/lnet/klnds/o2iblnd/o2iblnd.c
+++ b/lnet/klnds/o2iblnd/o2iblnd.c
@@ -2888,16 +2888,16 @@ kiblnd_create_dev(char *ifname)

                flags = dev_get_flags(netdev);
                if (!(flags & IFF_UP)) {
-                       CERROR("Can't query IPoIB interface %s: it's down\n",
+                       CWARN("Can't query IPoIB interface %s: it's down\n",
                               netdev->name);
-                       goto unlock;
+                       continue;
                }

                in_dev = __in_dev_get_rtnl(netdev);
                if (!in_dev) {
                        CWARN("Interface %s has no IPv4 status.\n",
                              netdev->name);
-                       goto unlock;
+                       continue;
                }

                for_ifa(in_dev)
@@ -2916,6 +2916,7 @@ kiblnd_create_dev(char *ifname)
                        }
                endfor_ifa(in_dev);
        }
+unlock:
        rtnl_unlock();

        if (!dev) {
@@ -2938,8 +2939,7 @@ kiblnd_create_dev(char *ifname)

        list_add_tail(&dev->ibd_list, &kiblnd_data.kib_devs);
        return dev;
-unlock:
-       rtnl_unlock();
+
 free_dev:
        LIBCFS_FREE(dev, sizeof(*dev));
        return NULL;

Issue Links

• duplicates: LU-12381 o2iblnd uses wrong IB interface