From 81a1c4cdcfe5a20edf760fcb857e9d5297c322dd Mon Sep 17 00:00:00 2001
From: Pranith Kumar K <pkarampu@redhat.com>
Date: Thu, 4 Aug 2016 00:41:16 +0530
Subject: [PATCH 107/141] cluster/ec: Do multi-threaded self-heal
>BUG: 1368451
>Change-Id: I5d6b91d714ad6906dc478a401e614115c89a8fbb
>Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
>Reviewed-on: http://review.gluster.org/15083
>Smoke: Gluster Build System <jenkins@build.gluster.org>
>Reviewed-by: Ashish Pandey <aspandey@redhat.com>
>NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
>CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
BUG: 1375465
Change-Id: Iad998614f659fd988fe41a2103cfc4af993aa8a2
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/87213
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
Tested-by: Atin Mukherjee <amukherj@redhat.com>
---
xlators/cluster/afr/src/afr.c | 6 ++--
xlators/cluster/ec/src/ec-heald.c | 14 +++++++++++-
xlators/cluster/ec/src/ec-heald.h | 2 +
xlators/cluster/ec/src/ec.c | 25 +++++++++++++++++++++++
xlators/mgmt/glusterd/src/glusterd-volume-set.c | 10 +++++++++
5 files changed, 52 insertions(+), 5 deletions(-)
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index d38d59c..c30e7ec 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -896,9 +896,9 @@ struct volume_options options[] = {
.min = 1,
.max = 64,
.default_value = "1",
- .description = "Maximum number of threads SHD can use per local "
- "brick. This can substantially lower heal times, "
- "but can also crush your bricks if you don't have "
+ .description = "Maximum number of parallel heals SHD can do per "
+ "local brick. This can substantially lower heal times"
+ ", but can also crush your bricks if you don't have "
"the storage hardware to support this."
},
{ .key = {"shd-wait-qlength"},
diff --git a/xlators/cluster/ec/src/ec-heald.c b/xlators/cluster/ec/src/ec-heald.c
index 0f63ab1..9860f10 100644
--- a/xlators/cluster/ec/src/ec-heald.c
+++ b/xlators/cluster/ec/src/ec-heald.c
@@ -240,6 +240,7 @@ ec_shd_index_sweep (struct subvol_healer *healer)
ec_t *ec = NULL;
int ret = 0;
xlator_t *subvol = NULL;
+ dict_t *xdata = NULL;
ec = healer->this->private;
subvol = ec->xl_list[healer->subvol];
@@ -252,9 +253,18 @@ ec_shd_index_sweep (struct subvol_healer *healer)
goto out;
}
- ret = syncop_dir_scan (subvol, &loc, GF_CLIENT_PID_SELF_HEALD,
- healer, ec_shd_index_heal);
+ xdata = dict_new ();
+ if (!xdata || dict_set_int32 (xdata, "get-gfid-type", 1)) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ ret = syncop_mt_dir_scan (NULL, subvol, &loc, GF_CLIENT_PID_SELF_HEALD,
+ healer, ec_shd_index_heal, xdata,
+ ec->shd.max_threads, ec->shd.wait_qlength);
out:
+ if (xdata)
+ dict_unref (xdata);
loc_wipe (&loc);
return ret;
diff --git a/xlators/cluster/ec/src/ec-heald.h b/xlators/cluster/ec/src/ec-heald.h
index 0f27a8e..0929044 100644
--- a/xlators/cluster/ec/src/ec-heald.h
+++ b/xlators/cluster/ec/src/ec-heald.h
@@ -34,6 +34,8 @@ struct _ec_self_heald {
gf_boolean_t iamshd;
gf_boolean_t enabled;
int timeout;
+ uint32_t max_threads;
+ uint32_t wait_qlength;
struct subvol_healer *index_healers;
struct subvol_healer *full_healers;
};
diff --git a/xlators/cluster/ec/src/ec.c b/xlators/cluster/ec/src/ec.c
index 7e295ca..587ea63 100644
--- a/xlators/cluster/ec/src/ec.c
+++ b/xlators/cluster/ec/src/ec.c
@@ -275,6 +275,10 @@ reconfigure (xlator_t *this, dict_t *options)
GF_OPTION_RECONF ("read-policy", read_policy, options, str, failed);
if (ec_assign_read_policy (ec, read_policy))
goto failed;
+ GF_OPTION_RECONF ("shd-max-threads", ec->shd.max_threads,
+ options, uint32, failed);
+ GF_OPTION_RECONF ("shd-wait-qlength", ec->shd.wait_qlength,
+ options, uint32, failed);
return 0;
failed:
@@ -616,6 +620,9 @@ init (xlator_t *this)
if (ec_assign_read_policy (ec, read_policy))
goto failed;
+ GF_OPTION_INIT ("shd-max-threads", ec->shd.max_threads, uint32, failed);
+ GF_OPTION_INIT ("shd-wait-qlength", ec->shd.wait_qlength, uint32, failed);
+
this->itable = inode_table_new (EC_SHD_INODE_LRU_LIMIT, this);
if (!this->itable)
goto failed;
@@ -1365,5 +1372,23 @@ struct volume_options options[] =
" subvolume using round-robin algo. 'gfid-hash' selects read"
" subvolume based on hash of the gfid of that file/directory.",
},
+ { .key = {"shd-max-threads"},
+ .type = GF_OPTION_TYPE_INT,
+ .min = 1,
+ .max = 64,
+ .default_value = "1",
+ .description = "Maximum number of parallel heals SHD can do per local "
+ "brick. This can substantially lower heal times, "
+ "but can also crush your bricks if you don't have "
+ "the storage hardware to support this."
+ },
+ { .key = {"shd-wait-qlength"},
+ .type = GF_OPTION_TYPE_INT,
+ .min = 1,
+ .max = 655536,
+ .default_value = "1024",
+ .description = "This option can be used to control number of heals"
+ " that can wait in SHD per subvolume"
+ },
{ }
};
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
index bd9b21b..8978709 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -2977,6 +2977,16 @@ struct volopt_map_entry glusterd_volopt_map[] = {
.op_version = GD_OP_VERSION_3_8_4,
.flags = OPT_FLAG_CLIENT_OPT
},
+ { .key = "disperse.shd-max-threads",
+ .voltype = "cluster/disperse",
+ .op_version = GD_OP_VERSION_3_9_0,
+ .flags = OPT_FLAG_CLIENT_OPT
+ },
+ { .key = "disperse.shd-wait-qlength",
+ .voltype = "cluster/disperse",
+ .op_version = GD_OP_VERSION_3_9_0,
+ .flags = OPT_FLAG_CLIENT_OPT
+ },
{ .key = NULL
}
};
--
1.7.1