From: Alex Zhuravlev <bzzz@whamcloud.com>
Date: Sun, 5 Nov 2023 13:51:29 +0000 (+0300)
Subject: LU-17261 lov: ignore broken components
X-Git-Tag: 2.15.60~37
X-Git-Url: https://git.whamcloud.com/?a=commitdiff_plain;h=902fe290e51dccdee89380fb725ae6e3c1802e2b;p=fs%2Flustre-release.git

LU-17261 lov: ignore broken components

if some component of a mirrored file is broken, it makes sense
to try another (possible valid) replica rather than give up
immediately.

Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: I32ea0efa90109f5159bf8b6c4e0efe1d543580c3
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/52996
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Zhenyu Xu <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
---

diff --git a/lustre/include/obd_support.h b/lustre/include/obd_support.h
index 0a42aa7..465382d 100644
--- a/lustre/include/obd_support.h
+++ b/lustre/include/obd_support.h
@@ -633,6 +633,7 @@ extern bool obd_enable_health_write;
 #define OBD_FAIL_LOV_MIRROR_INIT		    0x1425
 #define OBD_FAIL_LOV_COMP_MAGIC			    0x1426
 #define OBD_FAIL_LOV_COMP_PATTERN		    0x1427
+#define OBD_FAIL_LOV_INVALID_OSTIDX		    0x1428
 
 #define OBD_FAIL_FID_INDIR	0x1501
 #define OBD_FAIL_FID_INLMA	0x1502
diff --git a/lustre/lod/lod_lov.c b/lustre/lod/lod_lov.c
index 4256191..d7000b6 100644
--- a/lustre/lod/lod_lov.c
+++ b/lustre/lod/lod_lov.c
@@ -846,6 +846,13 @@ static int lod_gen_component_ea(const struct lu_env *env,
 		 * component, its l_ost_idx does not matter.
 		 */
 		objs[i].l_ost_idx = cpu_to_le32(ost_idx);
+
+		/* simulation of broken LOVEA */
+		if (CFS_FAIL_CHECK(OBD_FAIL_LOV_INVALID_OSTIDX) &&
+		    comp_idx == 0 && i == 0 && lo->ldo_mirror_count > 1) {
+			objs[i].l_ost_idx = cpu_to_le32(0xffffffff);
+		}
+
 	}
 done:
 	if (lmm_size != NULL)
diff --git a/lustre/lov/lov_ea.c b/lustre/lov/lov_ea.c
index c268197..62656c3 100644
--- a/lustre/lov/lov_ea.c
+++ b/lustre/lov/lov_ea.c
@@ -554,8 +554,15 @@ lsm_unpackmd_comp_md_v1(struct lov_obd *lov, void *buf, size_t buf_size)
 					LCME_FL_INIT,
 					(i == entry_count - 1) ? &maxbytes :
 								 NULL);
-		if (IS_ERR(lsme))
-			GOTO(out_lsm, rc = PTR_ERR(lsme));
+		if (IS_ERR(lsme)) {
+			OBD_ALLOC_LARGE(lsme, sizeof(*lsme));
+			if (!lsme)
+				GOTO(out_lsm, rc = -ENOMEM);
+
+			lsme->lsme_magic = LOV_MAGIC_FOREIGN;
+			lsme->lsme_pattern = LOV_PATTERN_FOREIGN;
+			lsme->lsme_flags = LCME_FL_OFFLINE;
+		}
 
 		/**
 		 * pressume that unrecognized magic component also has valid
diff --git a/lustre/tests/sanity-flr.sh b/lustre/tests/sanity-flr.sh
index 01f2b00..6fe457d 100644
--- a/lustre/tests/sanity-flr.sh
+++ b/lustre/tests/sanity-flr.sh
@@ -4435,6 +4435,19 @@ test_209b() {
 }
 run_test 209b "pagecache can be used after LL cancellation"
 
+test_210a() {
+	local tf=$DIR/$tfile
+
+	stack_trap "rm -f $tf"
+	dd if=/dev/zero of=$tf bs=1M count=1 || error "can't dd"
+#define OBD_FAIL_LOV_INVALID_OSTIDX		    0x1428
+	do_facet mds1 "$LCTL set_param fail_loc=0x1428"
+	$LFS mirror extend -N $tf || error "can't mirror"
+	$LFS getstripe -v $tf
+	stat $tf || error "can't stat"
+}
+run_test 210a "handle broken mirrored lovea"
+
 complete_test $SECONDS
 check_and_cleanup_lustre
 exit_status