From 107e5934f11a648bc06a87949f46e34ae1616f92 Mon Sep 17 00:00:00 2001
From: Krutika Dhananjay <kdhananj@redhat.com>
Date: Thu, 22 Sep 2016 16:48:54 +0530
Subject: [PATCH 201/206] cluster/afr: CLI for granular entry heal
enablement/disablement
Backport of: http://review.gluster.org/15747
When there are already existing non-granular indices created that are
yet to be healed, if granular-entry-heal option is toggled from 'off' to
'on', AFR self-heal whenever it kicks in, will try to look for granular
indices in 'entry-changes'. Because of the absence of name indices,
granular entry healing logic will fail to heal these directories, and
worse yet unset pending extended attributes with the assumption that
are no entries that need heal.
To get around this, a new CLI is introduced which will invoke glfsheal
program to figure whether at the time an attempt is made to enable
granular entry heal, there are pending heals on the volume OR there
are one or more bricks that are down. If either of them is true, the
command will be failed with the appropriate error.
New CLI: gluster volume heal <VOL> granular-entry-heal {enable,disable}
Change-Id: Idd9e044ac37143c4914e49d52ea427aa986c8291
BUG: 1385474
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/91374
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Tested-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
---
cli/src/cli-cmd-parser.c | 16 +-
cli/src/cli-cmd-volume.c | 28 ++--
cli/src/cli-rpc-ops.c | 8 +
heal/src/glfs-heal.c | 180 +++++++++++++++++----
rpc/rpc-lib/src/protocol-common.h | 2 +
tests/basic/afr/granular-esh/add-brick.t | 2 +-
tests/basic/afr/granular-esh/cli.t | 142 ++++++++++++++++
tests/basic/afr/granular-esh/conservative-merge.t | 4 +-
tests/basic/afr/granular-esh/granular-esh.t | 2 +-
.../granular-indices-but-non-granular-heal.t | 4 +-
tests/basic/afr/granular-esh/replace-brick.t | 2 +-
xlators/mgmt/glusterd/src/glusterd-op-sm.c | 19 +++
xlators/mgmt/glusterd/src/glusterd-volume-ops.c | 61 +++++--
13 files changed, 407 insertions(+), 63 deletions(-)
create mode 100644 tests/basic/afr/granular-esh/cli.t
diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index b419915..a40dcb1 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -3841,7 +3841,8 @@ cli_cmd_volume_heal_options_parse (const char **words, int wordcount,
if (wordcount == 5) {
if (strcmp (words[3], "info") &&
- strcmp (words[3], "statistics")) {
+ strcmp (words[3], "statistics") &&
+ strcmp (words[3], "granular-entry-heal")) {
ret = -1;
goto out;
}
@@ -3871,6 +3872,19 @@ cli_cmd_volume_heal_options_parse (const char **words, int wordcount,
goto done;
}
}
+
+ if (!strcmp (words[3], "granular-entry-heal")) {
+ if (!strcmp (words[4], "enable")) {
+ ret = dict_set_int32 (dict, "heal-op",
+ GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE);
+ goto done;
+ } else if (!strcmp (words[4], "disable")) {
+ ret = dict_set_int32 (dict, "heal-op",
+ GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE);
+ goto done;
+ }
+ }
+
ret = -1;
goto out;
}
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index b4fdecd..04fe432 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -2763,7 +2763,8 @@ cli_print_brick_status (cli_volume_status_t *status)
(op == GF_SHD_OP_SBRAIN_HEAL_FROM_LATEST_MTIME) ||\
(op == GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK) || \
(op == GF_SHD_OP_INDEX_SUMMARY) || \
- (op == GF_SHD_OP_SPLIT_BRAIN_FILES))
+ (op == GF_SHD_OP_SPLIT_BRAIN_FILES) || \
+ (op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE))
int
cli_launch_glfs_heal (int heal_op, dict_t *options)
@@ -2812,6 +2813,10 @@ cli_launch_glfs_heal (int heal_op, dict_t *options)
runner_add_args (&runner, "xml", NULL);
}
break;
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE:
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE:
+ runner_add_args (&runner, "granular-entry-heal-op", NULL);
+ break;
default:
ret = -1;
}
@@ -2823,11 +2828,11 @@ cli_launch_glfs_heal (int heal_op, dict_t *options)
printf ("%s", out);
}
ret = runner_end (&runner);
- ret = WEXITSTATUS (ret);
out:
return ret;
}
+
int
cli_cmd_volume_heal_cbk (struct cli_state *state, struct cli_cmd_word *word,
const char **words, int wordcount)
@@ -2864,19 +2869,19 @@ cli_cmd_volume_heal_cbk (struct cli_state *state, struct cli_cmd_word *word,
goto out;
if (NEEDS_GLFS_HEAL (heal_op)) {
ret = cli_launch_glfs_heal (heal_op, options);
- if (ret == -1)
+ if (ret < 0)
+ goto out;
+ if (heal_op != GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE)
goto out;
}
- else {
- proc = &cli_rpc_prog->proctable[GLUSTER_CLI_HEAL_VOLUME];
- CLI_LOCAL_INIT (local, words, frame, options);
+ proc = &cli_rpc_prog->proctable[GLUSTER_CLI_HEAL_VOLUME];
- if (proc->fn) {
- ret = proc->fn (frame, THIS, options);
- }
- }
+ CLI_LOCAL_INIT (local, words, frame, options);
+ if (proc->fn) {
+ ret = proc->fn (frame, THIS, options);
+ }
out:
if (ret) {
cli_cmd_sent_status_get (&sent);
@@ -3285,7 +3290,8 @@ struct cli_cmd volume_cmds[] = {
"statistics [heal-count [replica <HOSTNAME:BRICKNAME>]] |"
"info [healed | heal-failed | split-brain] |"
"split-brain {bigger-file <FILE> | latest-mtime <FILE> |"
- "source-brick <HOSTNAME:BRICKNAME> [<FILE>]}]",
+ "source-brick <HOSTNAME:BRICKNAME> [<FILE>]} |"
+ "granular-entry-heal {enable | disable}]",
cli_cmd_volume_heal_cbk,
"self-heal commands on volume specified by <VOLNAME>"},
diff --git a/cli/src/cli-rpc-ops.c b/cli/src/cli-rpc-ops.c
index bbb5894..bea5437 100644
--- a/cli/src/cli-rpc-ops.c
+++ b/cli/src/cli-rpc-ops.c
@@ -8985,6 +8985,14 @@ gf_cli_heal_volume_cbk (struct rpc_req *req, struct iovec *iov,
operation = "";
heal_op_str = "Disable heal";
break;
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE:
+ operation = "";
+ heal_op_str = "Enable granular entry heal";
+ break;
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE:
+ operation = "";
+ heal_op_str = "Disable granular entry heal";
+ break;
}
if (rsp.op_ret) {
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index 2a85e75..94cb6b0 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -39,7 +39,7 @@ xmlDocPtr glfsh_doc = NULL;
ret = 0; \
} while (0) \
-typedef void (*print_status) (dict_t *, char *, uuid_t, uint64_t *,
+typedef int (*print_status) (dict_t *, char *, uuid_t, uint64_t *,
gf_boolean_t flag);
int glfsh_heal_splitbrain_file (glfs_t *fs, xlator_t *top_subvol,
@@ -65,6 +65,11 @@ int32_t is_xml;
"source-brick <HOSTNAME:BRICKNAME> [<FILE>] | "\
"split-brain-info]\n"
+typedef enum {
+ GLFSH_MODE_CONTINUE_ON_ERROR = 1,
+ GLFSH_MODE_EXIT_ON_FIRST_FAILURE,
+} glfsh_fail_mode_t;
+
int
glfsh_init ()
{
@@ -72,6 +77,30 @@ glfsh_init ()
}
int
+glfsh_end_op_granular_entry_heal (int op_ret, char *op_errstr)
+{
+ /* If error sting is available, give it higher precedence.*/
+
+ if (op_errstr) {
+ printf ("%s\n", op_errstr);
+ } else if (op_ret < 0) {
+ if (op_ret == -EAGAIN)
+ printf ("One or more entries need heal. Please execute "
+ "the command again after there are no entries "
+ "to be healed\n");
+ else if (op_ret == -ENOTCONN)
+ printf ("One or more bricks could be down. Please "
+ "execute the command again after bringing all "
+ "bricks online and finishing any pending "
+ "heals\n");
+ else
+ printf ("Command failed - %s. Please check the logs for"
+ " more details\n", strerror (-op_ret));
+ }
+ return 0;
+}
+
+int
glfsh_end (int op_ret, char *op_errstr)
{
if (op_errstr)
@@ -87,6 +116,12 @@ glfsh_print_hr_spb_status (char *path, uuid_t gfid, char *status)
}
void
+glfsh_no_print_hr_heal_status (char *path, uuid_t gfid, char *status)
+{
+ return;
+}
+
+void
glfsh_print_hr_heal_status (char *path, uuid_t gfid, char *status)
{
printf ("%s%s\n", path, status);
@@ -296,6 +331,12 @@ out:
}
int
+glfsh_no_print_hr_heal_op_status (int ret, uint64_t num_entries, char *fmt_str)
+{
+ return 0;
+}
+
+int
glfsh_print_hr_heal_op_status (int ret, uint64_t num_entries, char *fmt_str)
{
if (ret < 0 && num_entries == 0) {
@@ -422,7 +463,7 @@ glfsh_index_purge (xlator_t *subvol, inode_t *inode, char *name)
return ret;
}
-void
+int
glfsh_print_spb_status (dict_t *dict, char *path, uuid_t gfid,
uint64_t *num_entries, gf_boolean_t flag)
{
@@ -434,7 +475,7 @@ glfsh_print_spb_status (dict_t *dict, char *path, uuid_t gfid,
ret = dict_get_str (dict, "heal-info", &value);
if (ret)
- return;
+ return 0;
if (!strcmp (value, "split-brain")) {
split_b = _gf_true;
@@ -456,10 +497,10 @@ glfsh_print_spb_status (dict_t *dict, char *path, uuid_t gfid,
gfid, NULL);
}
}
- return;
+ return 0;
}
-void
+int
glfsh_print_heal_status (dict_t *dict, char *path, uuid_t gfid,
uint64_t *num_entries, gf_boolean_t ignore_dirty)
{
@@ -471,7 +512,7 @@ glfsh_print_heal_status (dict_t *dict, char *path, uuid_t gfid,
ret = dict_get_str (dict, "heal-info", &value);
if (ret || (!strcmp (value, "no-heal")))
- return;
+ return 0;
if (!strcmp (value, "heal")) {
ret = gf_asprintf (&status, " ");
@@ -514,7 +555,7 @@ out:
if (pending) {
GF_FREE (status);
status = NULL;
- return;
+ return 0;
}
}
if (ret == -1)
@@ -527,7 +568,21 @@ out:
status ? status : "");
GF_FREE (status);
- return;
+ return 0;
+}
+
+int
+glfsh_heal_status_boolean (dict_t *dict, char *path, uuid_t gfid,
+ uint64_t *num_entries, gf_boolean_t ignore_dirty)
+{
+ int ret = 0;
+ char *value = NULL;
+
+ ret = dict_get_str (dict, "heal-info", &value);
+ if ((!ret) && (!strcmp (value, "no-heal")))
+ return 0;
+ else
+ return -1;
}
static int
@@ -561,11 +616,12 @@ static int
glfsh_process_entries (xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
uint64_t *offset, uint64_t *num_entries,
print_status glfsh_print_status,
- gf_boolean_t ignore_dirty)
+ gf_boolean_t ignore_dirty, glfsh_fail_mode_t mode)
{
gf_dirent_t *entry = NULL;
gf_dirent_t *tmp = NULL;
int ret = 0;
+ int print_status = 0;
char *path = NULL;
uuid_t gfid = {0};
xlator_t *this = NULL;
@@ -591,8 +647,13 @@ glfsh_process_entries (xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
gf_uuid_copy (loc.gfid, gfid);
ret = syncop_getxattr (this, &loc, &dict, GF_HEAL_INFO, NULL,
NULL);
- if (ret)
- continue;
+ if (ret) {
+ if ((mode != GLFSH_MODE_CONTINUE_ON_ERROR) &&
+ (ret == -ENOTCONN))
+ goto out;
+ else
+ continue;
+ }
ret = syncop_gfid_to_path (this->itable, xl, gfid, &path);
@@ -601,11 +662,19 @@ glfsh_process_entries (xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
ret = 0;
continue;
}
- if (dict)
- glfsh_print_status (dict, path, gfid,
- num_entries, ignore_dirty);
+ if (dict) {
+ print_status = glfsh_print_status (dict, path, gfid,
+ num_entries,
+ ignore_dirty);
+ if ((print_status) &&
+ (mode != GLFSH_MODE_CONTINUE_ON_ERROR)) {
+ ret = -EAGAIN;
+ goto out;
+ }
+ }
}
ret = 0;
+out:
GF_FREE (path);
if (dict) {
dict_unref (dict);
@@ -620,17 +689,21 @@ glfsh_crawl_directory (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
dict_t *xattr_req, uint64_t *num_entries,
gf_boolean_t ignore)
{
- uint64_t offset = 0;
+ int ret = 0;
+ int heal_op = -1;
+ uint64_t offset = 0;
gf_dirent_t entries;
- int ret = 0;
gf_boolean_t free_entries = _gf_false;
- int heal_op = -1;
+ glfsh_fail_mode_t mode = GLFSH_MODE_CONTINUE_ON_ERROR;
INIT_LIST_HEAD (&entries.list);
ret = dict_get_int32 (xattr_req, "heal-op", &heal_op);
if (ret)
return ret;
+ if (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE)
+ mode = GLFSH_MODE_EXIT_ON_FIRST_FAILURE;
+
while (1) {
ret = syncop_readdir (readdir_xl, fd, 131072, offset, &entries,
NULL, NULL);
@@ -647,7 +720,7 @@ glfsh_crawl_directory (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
&entries, &offset,
num_entries,
glfsh_print_heal_status,
- ignore);
+ ignore, mode);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_SPLIT_BRAIN_FILES) {
@@ -655,13 +728,20 @@ glfsh_crawl_directory (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
&entries, &offset,
num_entries,
glfsh_print_spb_status,
- ignore);
+ ignore, mode);
if (ret < 0)
goto out;
} else if (heal_op == GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK) {
ret = glfsh_heal_entries (fs, top_subvol, rootloc,
&entries, &offset,
num_entries, xattr_req);
+ } else if (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE) {
+ ret = glfsh_process_entries (readdir_xl, fd, &entries,
+ &offset, num_entries,
+ glfsh_heal_status_boolean,
+ ignore, mode);
+ if (ret < 0)
+ goto out;
}
gf_dirent_free (&entries);
free_entries = _gf_false;
@@ -674,6 +754,12 @@ out:
}
static int
+glfsh_no_print_brick_from_xl (xlator_t *xl, loc_t *rootloc)
+{
+ return 0;
+}
+
+static int
glfsh_print_brick_from_xl (xlator_t *xl, loc_t *rootloc)
{
char *remote_host = NULL;
@@ -751,6 +837,13 @@ glfsh_print_pending_heals (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
if (ret)
goto out;
+ if ((!is_parent_replicate) &&
+ ((heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE) ||
+ (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE))) {
+ ret = 0;
+ goto out;
+ }
+
ret = glfsh_output->print_brick_from_xl (xl, rootloc);
if (ret < 0)
goto out;
@@ -758,6 +851,10 @@ glfsh_print_pending_heals (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
ret = glfsh_print_pending_heals_type (fs, top_subvol, rootloc, xl,
heal_op, xattr_req,
GF_XATTROP_INDEX_GFID, &count);
+
+ if (ret < 0 && heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE)
+ goto out;
+
total += count;
count = 0;
if (ret == -ENOTCONN)
@@ -838,14 +935,14 @@ out:
return NULL;
}
-
int
glfsh_gather_heal_info (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
gf_xl_afr_op_t heal_op)
{
- xlator_t *xl = NULL;
+ int ret = 0;
+ xlator_t *xl = NULL;
xlator_t *heal_xl = NULL;
- xlator_t *old_THIS = NULL;
+ xlator_t *old_THIS = NULL;
xl = top_subvol;
while (xl->next)
@@ -856,20 +953,28 @@ glfsh_gather_heal_info (glfs_t *fs, xlator_t *top_subvol, loc_t *rootloc,
if (heal_xl) {
old_THIS = THIS;
THIS = heal_xl;
- glfsh_print_pending_heals (fs, top_subvol,
- rootloc, xl,
- heal_op,
- !strcmp
- (heal_xl->type,
- "cluster/replicate"));
+ ret = glfsh_print_pending_heals (fs, top_subvol,
+ rootloc, xl,
+ heal_op,
+ !strcmp
+ (heal_xl->type,
+ "cluster/replicate"));
THIS = old_THIS;
+
+ if ((ret < 0) &&
+ (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE))
+ goto out;
}
}
xl = xl->prev;
}
- return 0;
+out:
+ if (heal_op != GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE)
+ ret = 0;
+
+ return ret;
}
int
@@ -1107,6 +1212,15 @@ glfsh_info_t glfsh_human_readable = {
.end = glfsh_end
};
+glfsh_info_t glfsh_no_print = {
+ .init = glfsh_init,
+ .print_brick_from_xl = glfsh_no_print_brick_from_xl,
+ .print_heal_op_status = glfsh_no_print_hr_heal_op_status,
+ .print_heal_status = glfsh_no_print_hr_heal_status,
+ .print_spb_status = glfsh_no_print_hr_heal_status,
+ .end = glfsh_end_op_granular_entry_heal
+};
+
#if (HAVE_LIB_XML)
glfsh_info_t glfsh_xml_output = {
.init = glfsh_xml_init,
@@ -1150,6 +1264,8 @@ main (int argc, char **argv)
} else if (!strcmp (argv[2], "xml")) {
heal_op = GF_SHD_OP_INDEX_SUMMARY;
is_xml = 1;
+ } else if (!strcmp (argv[2], "granular-entry-heal-op")) {
+ heal_op = GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE;
} else {
printf (USAGE_STR, argv[0]);
ret = -1;
@@ -1206,6 +1322,9 @@ main (int argc, char **argv)
}
+ if (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE)
+ glfsh_output = &glfsh_no_print;
+
ret = glfsh_output->init ();
if (ret)
exit (EXIT_FAILURE);
@@ -1282,6 +1401,7 @@ main (int argc, char **argv)
switch (heal_op) {
case GF_SHD_OP_INDEX_SUMMARY:
case GF_SHD_OP_SPLIT_BRAIN_FILES:
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE:
ret = glfsh_gather_heal_info (fs, top_subvol, &rootloc,
heal_op);
break;
@@ -1300,6 +1420,8 @@ main (int argc, char **argv)
}
glfsh_output->end (ret, NULL);
+ if (ret < 0)
+ ret = -ret;
loc_wipe (&rootloc);
glfs_subvol_done (fs, top_subvol);
cleanup (fs);
diff --git a/rpc/rpc-lib/src/protocol-common.h b/rpc/rpc-lib/src/protocol-common.h
index c7c3327..0fe7d1f 100644
--- a/rpc/rpc-lib/src/protocol-common.h
+++ b/rpc/rpc-lib/src/protocol-common.h
@@ -254,6 +254,8 @@ typedef enum {
GF_SHD_OP_HEAL_ENABLE,
GF_SHD_OP_HEAL_DISABLE,
GF_SHD_OP_SBRAIN_HEAL_FROM_LATEST_MTIME,
+ GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE,
+ GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE,
} gf_xl_afr_op_t ;
struct gf_gsync_detailed_status_ {
diff --git a/tests/basic/afr/granular-esh/add-brick.t b/tests/basic/afr/granular-esh/add-brick.t
index f3125d7..270cf1d 100644
--- a/tests/basic/afr/granular-esh/add-brick.t
+++ b/tests/basic/afr/granular-esh/add-brick.t
@@ -14,7 +14,7 @@ TEST $CLI volume set $V0 cluster.data-self-heal off
TEST $CLI volume set $V0 cluster.metadata-self-heal off
TEST $CLI volume set $V0 cluster.entry-self-heal off
TEST $CLI volume set $V0 self-heal-daemon off
-TEST $CLI volume set $V0 granular-entry-heal on
+TEST $CLI volume heal $V0 granular-entry-heal enable
TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0
diff --git a/tests/basic/afr/granular-esh/cli.t b/tests/basic/afr/granular-esh/cli.t
new file mode 100644
index 0000000..a655180
--- /dev/null
+++ b/tests/basic/afr/granular-esh/cli.t
@@ -0,0 +1,142 @@
+#!/bin/bash
+
+. $(dirname $0)/../../../include.rc
+. $(dirname $0)/../../../volume.rc
+. $(dirname $0)/../../../afr.rc
+
+cleanup
+
+TESTS_EXPECTED_IN_LOOP=4
+
+TEST glusterd
+TEST pidof glusterd
+
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
+# Test that enabling the option should work on a newly created volume
+TEST $CLI volume set $V0 cluster.granular-entry-heal on
+TEST $CLI volume set $V0 cluster.granular-entry-heal off
+
+#########################
+##### DISPERSE TEST #####
+#########################
+# Execute the same command on a disperse volume and make sure it fails.
+TEST $CLI volume create $V1 disperse 3 redundancy 1 $H0:$B0/${V1}{0,1,2}
+TEST $CLI volume start $V1
+TEST ! $CLI volume heal $V1 granular-entry-heal enable
+TEST ! $CLI volume heal $V1 granular-entry-heal disable
+
+#######################
+###### TIER TEST ######
+#######################
+# Execute the same command on a disperse + replicate tiered volume and make
+# sure the option is set on the replicate leg of the volume
+TEST $CLI volume attach-tier $V1 replica 2 $H0:$B0/${V1}{3,4}
+TEST $CLI volume heal $V1 granular-entry-heal enable
+EXPECT "enable" volume_get_field $V1 cluster.granular-entry-heal
+TEST $CLI volume heal $V1 granular-entry-heal disable
+EXPECT "disable" volume_get_field $V1 cluster.granular-entry-heal
+
+# Kill a disperse brick and make heal be pending on the volume.
+TEST kill_brick $V1 $H0 $B0/${V1}0
+
+# Now make sure that one offline brick in disperse does not affect enabling the
+# option on the volume.
+TEST $CLI volume heal $V1 granular-entry-heal enable
+EXPECT "enable" volume_get_field $V1 cluster.granular-entry-heal
+TEST $CLI volume heal $V1 granular-entry-heal disable
+EXPECT "disable" volume_get_field $V1 cluster.granular-entry-heal
+
+# Now kill a replicate brick.
+TEST kill_brick $V1 $H0 $B0/${V1}3
+# Now make sure that one offline brick in replicate causes the command to be
+# failed.
+TEST ! $CLI volume heal $V1 granular-entry-heal enable
+EXPECT "disable" volume_get_field $V1 cluster.granular-entry-heal
+
+######################
+### REPLICATE TEST ###
+######################
+TEST $CLI volume start $V0
+TEST $CLI volume set $V0 cluster.data-self-heal off
+TEST $CLI volume set $V0 cluster.metadata-self-heal off
+TEST $CLI volume set $V0 cluster.entry-self-heal off
+TEST $CLI volume set $V0 self-heal-daemon off
+# Test that the volume-set way of enabling the option is disallowed
+TEST ! $CLI volume set $V0 granular-entry-heal on
+# Test that the volume-heal way of enabling the option is allowed
+TEST $CLI volume heal $V0 granular-entry-heal enable
+# Volume-reset of the option should be allowed
+TEST $CLI volume reset $V0 granular-entry-heal
+TEST $CLI volume heal $V0 granular-entry-heal enable
+
+EXPECT "enable" volume_option $V0 cluster.granular-entry-heal
+
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0
+
+# Kill brick-0.
+TEST kill_brick $V0 $H0 $B0/${V0}0
+
+# Disabling the option should work even when one or more bricks are down
+TEST $CLI volume heal $V0 granular-entry-heal disable
+# When a brick is down, 'enable' attempt should be failed
+TEST ! $CLI volume heal $V0 granular-entry-heal enable
+
+# Restart the killed brick
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+
+# When all bricks are up, it should be possible to enable the option
+TEST $CLI volume heal $V0 granular-entry-heal enable
+
+# Kill brick-0 again
+TEST kill_brick $V0 $H0 $B0/${V0}0
+
+# Create files under root
+for i in {1..2}
+do
+ echo $i > $M0/f$i
+done
+
+# Test that the index associated with '/' is created on B1.
+TEST stat $B0/${V0}1/.glusterfs/indices/entry-changes/$ROOT_GFID
+
+# Check for successful creation of granular entry indices
+for i in {1..2}
+do
+ TEST_IN_LOOP stat $B0/${V0}1/.glusterfs/indices/entry-changes/$ROOT_GFID/f$i
+done
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+
+TEST gluster volume set $V0 cluster.self-heal-daemon on
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
+
+TEST $CLI volume heal $V0
+
+# Wait for heal to complete
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+
+# Test if data was healed
+for i in {1..2}
+do
+ TEST_IN_LOOP diff $B0/${V0}0/f$i $B0/${V0}1/f$i
+done
+
+# Now verify that there are no name indices left after self-heal
+TEST ! stat $B0/${V0}1/.glusterfs/indices/entry-changes/$ROOT_GFID/f1
+TEST ! stat $B0/${V0}1/.glusterfs/indices/entry-changes/$ROOT_GFID/f2
+TEST ! stat $B0/${V0}1/.glusterfs/indices/entry-changes/$ROOT_GFID
+
+# Perform a volume-reset-all-options operation
+TEST $CLI volume reset $V0
+# Ensure that granular entry heal is also disabled
+EXPECT "no" volume_get_field $V0 cluster.granular-entry-heal
+EXPECT "on" volume_get_field $V0 cluster.entry-self-heal
+
+cleanup
+#G_TESTDEF_TEST_STATUS_NETBSD7=BAD_TEST,BUG=1399038
diff --git a/tests/basic/afr/granular-esh/conservative-merge.t b/tests/basic/afr/granular-esh/conservative-merge.t
index b566a0e..b170e47 100644
--- a/tests/basic/afr/granular-esh/conservative-merge.t
+++ b/tests/basic/afr/granular-esh/conservative-merge.t
@@ -11,13 +11,13 @@ TESTS_EXPECTED_IN_LOOP=4
TEST glusterd
TEST pidof glusterd
TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
+TEST $CLI volume start $V0
TEST $CLI volume set $V0 self-heal-daemon off
TEST $CLI volume set $V0 data-self-heal off
TEST $CLI volume set $V0 metadata-self-heal off
TEST $CLI volume set $V0 entry-self-heal off
-TEST $CLI volume set $V0 granular-entry-heal on
+TEST $CLI volume heal $V0 granular-entry-heal enable
-TEST $CLI volume start $V0
TEST $GFS --volfile-id=$V0 -s $H0 $M0
TEST mkdir $M0/dir
diff --git a/tests/basic/afr/granular-esh/granular-esh.t b/tests/basic/afr/granular-esh/granular-esh.t
index ee53878..de0e8f4 100644
--- a/tests/basic/afr/granular-esh/granular-esh.t
+++ b/tests/basic/afr/granular-esh/granular-esh.t
@@ -16,7 +16,7 @@ TEST $CLI volume set $V0 cluster.data-self-heal off
TEST $CLI volume set $V0 cluster.metadata-self-heal off
TEST $CLI volume set $V0 cluster.entry-self-heal off
TEST $CLI volume set $V0 self-heal-daemon off
-TEST $CLI volume set $V0 granular-entry-heal on
+TEST $CLI volume heal $V0 granular-entry-heal enable
TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0
diff --git a/tests/basic/afr/granular-esh/granular-indices-but-non-granular-heal.t b/tests/basic/afr/granular-esh/granular-indices-but-non-granular-heal.t
index 2da90a9..1b5421b 100644
--- a/tests/basic/afr/granular-esh/granular-indices-but-non-granular-heal.t
+++ b/tests/basic/afr/granular-esh/granular-indices-but-non-granular-heal.t
@@ -17,7 +17,7 @@ TEST $CLI volume set $V0 cluster.data-self-heal off
TEST $CLI volume set $V0 cluster.metadata-self-heal off
TEST $CLI volume set $V0 cluster.entry-self-heal off
TEST $CLI volume set $V0 self-heal-daemon off
-TEST $CLI volume set $V0 granular-entry-heal on
+TEST $CLI volume heal $V0 granular-entry-heal enable
TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0
@@ -40,7 +40,7 @@ do
done
# Now disable granular-entry-heal
-TEST $CLI volume set $V0 granular-entry-heal off
+TEST $CLI volume heal $V0 granular-entry-heal disable
# Start the brick that was down
TEST $CLI volume start $V0 force
diff --git a/tests/basic/afr/granular-esh/replace-brick.t b/tests/basic/afr/granular-esh/replace-brick.t
index aaa54da..639ed81 100644
--- a/tests/basic/afr/granular-esh/replace-brick.t
+++ b/tests/basic/afr/granular-esh/replace-brick.t
@@ -12,7 +12,7 @@ TEST $CLI volume set $V0 cluster.data-self-heal off
TEST $CLI volume set $V0 cluster.metadata-self-heal off
TEST $CLI volume set $V0 cluster.entry-self-heal off
TEST $CLI volume set $V0 self-heal-daemon off
-TEST $CLI volume set $V0 granular-entry-heal on
+TEST $CLI volume heal $V0 granular-entry-heal enable
TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0;
diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index 67f5f21..9afeb52 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -1120,6 +1120,25 @@ glusterd_op_stage_set_volume (dict_t *dict, char **op_errstr)
if (key_fixed)
key = key_fixed;
+ if (strcmp (key, "cluster.granular-entry-heal") == 0) {
+ /* For granular entry-heal, if the set command was
+ * invoked through volume-set CLI, then allow the
+ * command only if the volume is still in 'Created'
+ * state
+ */
+ if ((dict_get (dict, "is-special-key") == NULL) &&
+ (volinfo->status != GLUSTERD_STATUS_NONE)) {
+ snprintf (errstr, sizeof (errstr), " 'gluster "
+ "volume set <VOLNAME> %s {enable, "
+ "disable}' is not supported. Use "
+ "'gluster volume heal <VOLNAME> "
+ "granular-entry-heal {enable, "
+ "disable}' instead.", key);
+ ret = -1;
+ goto out;
+ }
+ }
+
/* Check if the key is cluster.op-version and set
* local_new_op_version to the value given if possible.
*/
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
index 95dbafa..35336e2 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
@@ -769,8 +769,9 @@ out:
return ret;
}
static int
-glusterd_handle_heal_enable_disable (rpcsvc_request_t *req, dict_t *dict,
- glusterd_volinfo_t *volinfo)
+glusterd_handle_heal_options_enable_disable (rpcsvc_request_t *req,
+ dict_t *dict,
+ glusterd_volinfo_t *volinfo)
{
gf_xl_afr_op_t heal_op = GF_SHD_OP_INVALID;
int ret = 0;
@@ -784,30 +785,58 @@ glusterd_handle_heal_enable_disable (rpcsvc_request_t *req, dict_t *dict,
}
if ((heal_op != GF_SHD_OP_HEAL_ENABLE) &&
- (heal_op != GF_SHD_OP_HEAL_DISABLE)) {
+ (heal_op != GF_SHD_OP_HEAL_DISABLE) &&
+ (heal_op != GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE) &&
+ (heal_op != GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE)) {
ret = -EINVAL;
goto out;
}
- if (heal_op == GF_SHD_OP_HEAL_ENABLE) {
+ if (((heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE) ||
+ (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE)) &&
+ (volinfo->type == GF_CLUSTER_TYPE_DISPERSE)) {
+ ret = -1;
+ goto out;
+ }
+
+ if ((heal_op == GF_SHD_OP_HEAL_ENABLE) ||
+ (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE)) {
value = "enable";
- } else if (heal_op == GF_SHD_OP_HEAL_DISABLE) {
+ } else if ((heal_op == GF_SHD_OP_HEAL_DISABLE) ||
+ (heal_op == GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE)) {
value = "disable";
}
/* Convert this command to volume-set command based on volume type */
if (volinfo->type == GF_CLUSTER_TYPE_TIER) {
- ret = glusterd_handle_shd_option_for_tier (volinfo, value,
- dict);
- if (!ret)
- goto set_volume;
- goto out;
+ switch (heal_op) {
+ case GF_SHD_OP_HEAL_ENABLE:
+ case GF_SHD_OP_HEAL_DISABLE:
+ ret = glusterd_handle_shd_option_for_tier (volinfo,
+ value, dict);
+ if (!ret)
+ goto set_volume;
+ goto out;
+ /* For any other heal_op, including granular-entry heal,
+ * just break out of the block but don't goto out yet.
+ */
+ default:
+ break;
+ }
}
- key = volgen_get_shd_key (volinfo->type);
- if (!key) {
- ret = -1;
- goto out;
+ if ((heal_op == GF_SHD_OP_HEAL_ENABLE) ||
+ (heal_op == GF_SHD_OP_HEAL_DISABLE)) {
+ key = volgen_get_shd_key (volinfo->type);
+ if (!key) {
+ ret = -1;
+ goto out;
+ }
+ } else {
+ key = "cluster.granular-entry-heal";
+ ret = dict_set_int8 (dict, "is-special-key", 1);
+ if (ret)
+ goto out;
}
ret = dict_set_str (dict, "key1", key);
@@ -893,7 +922,7 @@ __glusterd_handle_cli_heal_volume (rpcsvc_request_t *req)
goto out;
}
- ret = glusterd_handle_heal_enable_disable (req, dict, volinfo);
+ ret = glusterd_handle_heal_options_enable_disable (req, dict, volinfo);
if (ret == -EINVAL) {
ret = 0;
} else {
@@ -1832,6 +1861,8 @@ glusterd_handle_heal_cmd (xlator_t *this, glusterd_volinfo_t *volinfo,
case GF_SHD_OP_INVALID:
case GF_SHD_OP_HEAL_ENABLE: /* This op should be handled in volume-set*/
case GF_SHD_OP_HEAL_DISABLE:/* This op should be handled in volume-set*/
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_ENABLE: /* This op should be handled in volume-set */
+ case GF_SHD_OP_GRANULAR_ENTRY_HEAL_DISABLE: /* This op should be handled in volume-set */
case GF_SHD_OP_SBRAIN_HEAL_FROM_BIGGER_FILE:/*glfsheal cmd*/
case GF_SHD_OP_SBRAIN_HEAL_FROM_LATEST_MTIME:/*glfsheal cmd*/
case GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK:/*glfsheal cmd*/
--
2.9.3