From d4aaa00d77d7a10748a1ca9af82265ad9ac42598 Mon Sep 17 00:00:00 2001 From: Joseph Fernandes Date: Sun, 8 Nov 2015 15:33:15 +0530 Subject: tier/ctr: Providing option to record or ignore metadata heat Currently we heat up a file for both data and metadata write. Here we provide a ctr xlator option called "ctr-record-metadata-heat" were the admin can decide on recording metadata heat i.e heatup a file on metadata writes or not. Metadata data operation are a. setattr: explicit changing of atime/mtime using utimes, changing of posix permissions of the file b. rename: Renaming a file, c. unlink, link: adding or deleting hardlinks d. xattrs: setting or removal of xattrs. NOTE: atime, mtime and ctime change through writev, readv, truncate, mknod and create will not be considered here as these fops are data and primary metadata fops. Defaultly "ctr-record-metadata-heat" is off. Admin can switch it on using gluster volume set command. Change-Id: I91157509255dd5cb429cda2b6d4f64582e155e7b BUG: 1279166 Signed-off-by: Joseph Fernandes Reviewed-on: http://review.gluster.org/12540 Tested-by: NetBSD Build System Tested-by: Gluster Build System Reviewed-by: Dan Lambright Tested-by: Dan Lambright --- libglusterfs/src/gfdb/gfdb_sqlite3_helper.c | 24 ++- tests/basic/tier/record-metadata-heat.t | 101 +++++++++ .../changetimerecorder/src/changetimerecorder.c | 229 ++++++++++++++++----- .../features/changetimerecorder/src/ctr-helper.c | 5 + .../features/changetimerecorder/src/ctr-helper.h | 55 ++++- xlators/mgmt/glusterd/src/glusterd-volume-set.c | 16 ++ 6 files changed, 373 insertions(+), 57 deletions(-) create mode 100755 tests/basic/tier/record-metadata-heat.t diff --git a/libglusterfs/src/gfdb/gfdb_sqlite3_helper.c b/libglusterfs/src/gfdb/gfdb_sqlite3_helper.c index 309d79cb965..17480e1ff4a 100644 --- a/libglusterfs/src/gfdb/gfdb_sqlite3_helper.c +++ b/libglusterfs/src/gfdb/gfdb_sqlite3_helper.c @@ -807,15 +807,33 @@ gf_sql_insert_wind (gf_sql_connection_t *sql_conn, gfdb_db_record->file_name, gfdb_db_record->file_path, gfdb_db_record->link_consistency, - gfdb_db_record->ignore_errors); + _gf_true); if (ret) { gf_msg (GFDB_STR_SQLITE3, - _gfdb_log_level (GF_LOG_ERROR, + _gfdb_log_level (GF_LOG_WARNING, gfdb_db_record->ignore_errors), 0, LG_MSG_INSERT_FAILED, "Failed " "inserting link in DB"); - goto out; + /* Even if link creation is failed we + * continue with the creation of file record. + * This covers to cases + * 1) Lookup heal: If the file record from + * gf_file_tb is deleted but the link record + * still exist. Lookup heal will attempt a heal + * with create_wind set. The link heal will fail + * as there is already a record and if we dont + * ignore the error we will not heal the + * gf_file_tb. + * 2) Rename file in cold tier: During a rename + * of a file that is there in cold tier. We get + * an link record created in hot tier for the + * linkto file. When the file gets heated and + * moves to hot tier there will be attempt from + * ctr lookup heal to create link and file + * record and If we dont ignore the error we + * will not heal the gf_file_tb. + * */ } gfdb_db_record->islinkupdate = gfdb_db_record-> link_consistency; diff --git a/tests/basic/tier/record-metadata-heat.t b/tests/basic/tier/record-metadata-heat.t new file mode 100755 index 00000000000..d36ed80591a --- /dev/null +++ b/tests/basic/tier/record-metadata-heat.t @@ -0,0 +1,101 @@ +#!/bin/bash + +. $(dirname $0)/../../include.rc +. $(dirname $0)/../../volume.rc +. $(dirname $0)/../../tier.rc + +NUM_BRICKS=3 +DEMOTE_FREQ=5 +DEMOTE_TIMEOUT=10 +PROMOTE_FREQ=5 + +FILE="file1.txt" +FILE_LINK="file2.txt" + +# Creates a tiered volume with pure distribute hot and cold tiers +# Both hot and cold tiers will have an equal number of bricks. + +function create_dist_tier_vol () { + mkdir $B0/cold + mkdir $B0/hot + TEST $CLI volume create $V0 $H0:$B0/cold/${V0}{0..$1} + TEST $CLI volume set $V0 performance.quick-read off + TEST $CLI volume set $V0 performance.io-cache off + TEST $CLI volume set $V0 features.ctr-enabled on + TEST $CLI volume start $V0 + TEST $CLI volume attach-tier $V0 $H0:$B0/hot/${V0}{0..$1} + TEST $CLI volume set $V0 cluster.tier-demote-frequency $DEMOTE_FREQ + TEST $CLI volume set $V0 cluster.tier-promote-frequency $PROMOTE_FREQ + TEST $CLI volume set $V0 cluster.read-freq-threshold 4 + TEST $CLI volume set $V0 cluster.write-freq-threshold 4 +} + + +cleanup; + +#Basic checks +TEST glusterd +TEST pidof glusterd +TEST $CLI volume info + + +#Create and start a tiered volume +create_dist_tier_vol $NUM_BRICKS + +# Mount FUSE +TEST glusterfs -s $H0 --volfile-id $V0 $M0 + + +# The file will be created on the hot tier +touch "$M0/$FILE" + +# Get the path of the file on the hot tier +HPATH=`find $B0/hot/ -name "$FILE"` +echo "File path on hot tier: "$HPATH + +# Expecting the file to be on the hot tier +EXPECT "yes" exists_and_regular_file $HPATH + +sleep_until_mid_cycle $DEMOTE_FREQ + +# Try to heat the file using 5 metadata operations +# WITHOUT setting ctr-record-metadata-heat on +touch "$M0/$FILE" +chmod +x "$M0/$FILE" +chown root "$M0/$FILE" +ln "$M0/$FILE" "$M0/$FILE_LINK" +rm -rf "$M0/$FILE_LINK" + +# Wait for the tier process to demote the file +sleep $DEMOTE_TIMEOUT + +# Get the path of the file on the cold tier +CPATH=`find $B0/cold/ -name "$FILE"` +echo "File path on cold tier: "$CPATH + +# Expecting the file to be on cold tier +EXPECT "yes" exists_and_regular_file $CPATH + +#Set ctr-record-metadata-heat on +TEST $CLI volume set $V0 ctr-record-metadata-heat on + +sleep_until_mid_cycle $DEMOTE_FREQ + +# Heating the file using 5 metadata operations +touch "$M0/$FILE" +chmod +x "$M0/$FILE" +chown root "$M0/$FILE" +ln "$M0/$FILE" "$M0/$FILE_LINK" +rm -rf "$M0/$FILE_LINK" + +# Wait for the tier process to demote the file +sleep $DEMOTE_TIMEOUT + +# Get the path of the file on the hot tier +echo "File path on hot tier: "$HPATH + +# Expecting the file to be on the hot tier +EXPECT "yes" exists_and_regular_file $HPATH + +cleanup; + diff --git a/xlators/features/changetimerecorder/src/changetimerecorder.c b/xlators/features/changetimerecorder/src/changetimerecorder.c index 2dd99dbb5cd..1831316f8a3 100644 --- a/xlators/features/changetimerecorder/src/changetimerecorder.c +++ b/xlators/features/changetimerecorder/src/changetimerecorder.c @@ -352,7 +352,7 @@ ctr_lookup (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_LINK_WIND_FAILED, - "Failed inserting link wind"); + "Failed to insert link wind"); } out: @@ -380,7 +380,7 @@ ctr_writev_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_WRITEV_UNWIND_FAILED, - "Failed inserting writev unwind"); + "Failed to insert writev unwind"); } @@ -414,7 +414,7 @@ ctr_writev (call_frame_t *frame, xlator_t *this, fd_t *fd, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_WRITEV_WIND_FAILED, - "Failed inserting writev wind"); + "Failed to insert writev wind"); } out: @@ -443,7 +443,7 @@ ctr_setattr_cbk (call_frame_t *frame, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_SETATTR_UNWIND_FAILED, - "Failed inserting setattr unwind"); + "Failed to insert setattr unwind"); } out: @@ -465,6 +465,7 @@ ctr_setattr (call_frame_t *frame, CTR_IS_DISABLED_THEN_GOTO(this, out); CTR_IF_INTERNAL_FOP_THEN_GOTO (frame, xdata, out); + CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO (this, out); /*Fill ctr inode context*/ FILL_CTR_INODE_CONTEXT(_inode_cx, loc->inode->ia_type, @@ -476,7 +477,7 @@ ctr_setattr (call_frame_t *frame, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_SETATTR_WIND_FAILED, - "Failed inserting setattr wind"); + "Failed to insert setattr wind"); } out: @@ -487,6 +488,65 @@ out: return 0; } +/*************************** fsetattr ***************************************/ +int32_t +ctr_fsetattr_cbk (call_frame_t *frame, + void *cookie, xlator_t *this, int32_t op_ret, + int32_t op_errno, struct iatt *preop_stbuf, + struct iatt *postop_stbuf, dict_t *xdata) +{ + int ret = -1; + + CTR_IS_DISABLED_THEN_GOTO(this, out); + + ret = ctr_insert_unwind(frame, this, + GFDB_FOP_INODE_WRITE, GFDB_FOP_UNWIND); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, + CTR_MSG_INSERT_SETATTR_UNWIND_FAILED, + "Failed to insert fsetattr unwind"); + } + +out: + STACK_UNWIND_STRICT (fsetattr, frame, op_ret, op_errno, + preop_stbuf, postop_stbuf, xdata); + + return 0; +} + + +int32_t +ctr_fsetattr (call_frame_t *frame, + xlator_t *this, fd_t *fd, + struct iatt *stbuf, int32_t valid, dict_t *xdata) +{ + int ret = -1; + gf_ctr_inode_context_t ctr_inode_cx; + gf_ctr_inode_context_t *_inode_cx = &ctr_inode_cx; + + CTR_IS_DISABLED_THEN_GOTO(this, out); + CTR_IF_INTERNAL_FOP_THEN_GOTO (frame, xdata, out); + CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO (this, out); + + /*Fill ctr inode context*/ + FILL_CTR_INODE_CONTEXT(_inode_cx, fd->inode->ia_type, + fd->inode->gfid, NULL, NULL, GFDB_FOP_INODE_WRITE, + GFDB_FOP_WIND); + + /*record into the database*/ + ret = ctr_insert_wind(frame, this, _inode_cx); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, + CTR_MSG_INSERT_SETATTR_WIND_FAILED, + "Failed to insert fsetattr wind"); + } +out: + STACK_WIND (frame, ctr_fsetattr_cbk, + FIRST_CHILD (this), FIRST_CHILD (this)->fops->fsetattr, + fd, stbuf, valid, xdata); + + return 0; +} /****************************fremovexattr************************************/ int32_t @@ -503,7 +563,7 @@ ctr_fremovexattr_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FREMOVEXATTR_UNWIND_FAILED, - "Failed inserting fremovexattr unwind"); + "Failed to insert fremovexattr unwind"); } out: @@ -522,7 +582,7 @@ ctr_fremovexattr (call_frame_t *frame, xlator_t *this, fd_t *fd, CTR_IS_DISABLED_THEN_GOTO(this, out); CTR_IF_INTERNAL_FOP_THEN_GOTO (frame, xdata, out); - + CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO (this, out); /*Fill ctr inode context*/ FILL_CTR_INODE_CONTEXT(_inode_cx, fd->inode->ia_type, @@ -534,7 +594,7 @@ ctr_fremovexattr (call_frame_t *frame, xlator_t *this, fd_t *fd, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FREMOVEXATTR_WIND_FAILED, - "Failed inserting fremovexattr wind"); + "Failed to insert fremovexattr wind"); } out: @@ -561,7 +621,7 @@ ctr_removexattr_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_REMOVEXATTR_UNWIND_FAILED, - "Failed inserting removexattr unwind"); + "Failed to insert removexattr unwind"); } out: @@ -580,7 +640,7 @@ ctr_removexattr (call_frame_t *frame, xlator_t *this, loc_t *loc, CTR_IS_DISABLED_THEN_GOTO(this, out); CTR_IF_INTERNAL_FOP_THEN_GOTO (frame, xdata, out); - + CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO (this, out); /*Fill ctr inode context*/ FILL_CTR_INODE_CONTEXT(_inode_cx, loc->inode->ia_type, @@ -592,7 +652,7 @@ ctr_removexattr (call_frame_t *frame, xlator_t *this, loc_t *loc, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_REMOVEXATTR_WIND_FAILED, - "Failed inserting removexattr wind"); + "Failed to insert removexattr wind"); } out: @@ -619,7 +679,7 @@ ctr_truncate_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_TRUNCATE_UNWIND_FAILED, - "Failed inserting truncate unwind"); + "Failed to insert truncate unwind"); } @@ -651,7 +711,7 @@ ctr_truncate (call_frame_t *frame, xlator_t *this, loc_t *loc, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_TRUNCATE_WIND_FAILED, - "Failed inserting truncate wind"); + "Failed to insert truncate wind"); } out: STACK_WIND (frame, ctr_truncate_cbk, FIRST_CHILD (this), @@ -676,7 +736,7 @@ ctr_ftruncate_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FTRUNCATE_UNWIND_FAILED, - "Failed inserting ftruncate unwind"); + "Failed to insert ftruncate unwind"); } out: @@ -707,7 +767,7 @@ ctr_ftruncate (call_frame_t *frame, xlator_t *this, fd_t *fd, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FTRUNCATE_WIND_FAILED, - "Failed inserting ftruncate wind"); + "Failed to insert ftruncate wind"); } out: @@ -735,7 +795,7 @@ ctr_rename_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_RENAME_UNWIND_FAILED, - "Failed inserting rename unwind"); + "Failed to insert rename unwind"); } out: @@ -774,12 +834,15 @@ ctr_rename (call_frame_t *frame, xlator_t *this, loc_t *oldloc, oldloc->inode->gfid, _nlink_cx, _olink_cx, GFDB_FOP_DENTRY_WRITE, GFDB_FOP_WIND); + /* Is a metatdata fop */ + _inode_cx->is_metadata_fop = _gf_true; + /*record into the database*/ ret = ctr_insert_wind(frame, this, _inode_cx); if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_RENAME_WIND_FAILED, - "Failed inserting rename wind"); + "Failed to insert rename wind"); } else { /* We are doing updation of hard link in inode context in wind * As we dont get the "inode" in the call back for rename */ @@ -834,7 +897,7 @@ ctr_unlink_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_UNLINK_UNWIND_FAILED, - "Failed inserting unlink unwind"); + "Failed to insert unlink unwind"); } } /*Last link that was deleted*/ @@ -845,7 +908,7 @@ ctr_unlink_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_UNLINK_UNWIND_FAILED, - "Failed inserting unlink unwind"); + "Failed to insert unlink unwind"); } } @@ -883,6 +946,9 @@ ctr_unlink (call_frame_t *frame, xlator_t *this, /*Internal FOP*/ _inode_cx->is_internal_fop = is_internal_fop (frame, xdata); + /* Is a metadata FOP */ + _inode_cx->is_metadata_fop = _gf_true; + /* If its a internal FOP and dht link file donot record*/ if (_inode_cx->is_internal_fop && dht_is_linkfile (&dummy_stat, xdata)) { @@ -894,7 +960,7 @@ ctr_unlink (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_UNLINK_UNWIND_FAILED, - "Failed inserting unlink wind"); + "Failed to insert unlink wind"); } else { /* We are doing delete of hard link in inode context in wind * As we dont get the "inode" in the call back for rename */ @@ -961,7 +1027,7 @@ ctr_fsync_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FSYNC_UNWIND_FAILED, - "Failed inserting fsync unwind"); + "Failed to insert fsync unwind"); } out: @@ -992,7 +1058,7 @@ ctr_fsync (call_frame_t *frame, xlator_t *this, fd_t *fd, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FSYNC_WIND_FAILED, - "Failed inserting fsync wind"); + "Failed to insert fsync wind"); } out: @@ -1017,10 +1083,9 @@ ctr_setxattr_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_FSYNC_UNWIND_FAILED, - "Failed inserting fsync unwind"); + "Failed to insert setxattr unwind"); } - out: STACK_UNWIND_STRICT (setxattr, frame, op_ret, op_errno, xdata); @@ -1037,6 +1102,7 @@ ctr_setxattr (call_frame_t *frame, xlator_t *this, CTR_IS_DISABLED_THEN_GOTO(this, out); CTR_IF_INTERNAL_FOP_THEN_GOTO (frame, xdata, out); + CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO (this, out); /*Fill ctr inode context*/ FILL_CTR_INODE_CONTEXT(_inode_cx, loc->inode->ia_type, @@ -1048,7 +1114,7 @@ ctr_setxattr (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_SETATTR_WIND_FAILED, - "Failed inserting setxattr wind"); + "Failed to insert setxattr wind"); } out: @@ -1057,7 +1123,61 @@ out: loc, xattr, flags, xdata); return 0; } +/**************************** fsetxattr *************************************/ +int32_t +ctr_fsetxattr_cbk (call_frame_t *frame, + void *cookie, xlator_t *this, int32_t op_ret, + int32_t op_errno, dict_t *xdata) +{ + int ret = -1; + + CTR_IS_DISABLED_THEN_GOTO(this, out); + + ret = ctr_insert_unwind(frame, this, GFDB_FOP_INODE_WRITE, + GFDB_FOP_UNWIND); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, + CTR_MSG_INSERT_FSYNC_UNWIND_FAILED, + "Failed to insert fsetxattr unwind"); + } + +out: + STACK_UNWIND_STRICT (fsetxattr, frame, op_ret, op_errno, xdata); + + return 0; +} + +int32_t +ctr_fsetxattr (call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *dict, + int32_t flags, dict_t *xdata) +{ + int ret = -1; + gf_ctr_inode_context_t ctr_inode_cx; + gf_ctr_inode_context_t *_inode_cx = &ctr_inode_cx; + + CTR_IS_DISABLED_THEN_GOTO(this, out); + CTR_IF_INTERNAL_FOP_THEN_GOTO (frame, xdata, out); + CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO (this, out); + + /*Fill ctr inode context*/ + FILL_CTR_INODE_CONTEXT(_inode_cx, fd->inode->ia_type, + fd->inode->gfid, NULL, NULL, + GFDB_FOP_INODE_WRITE, GFDB_FOP_WIND); + + /*record into the database*/ + ret = ctr_insert_wind(frame, this, _inode_cx); + if (ret) { + gf_msg (this->name, GF_LOG_ERROR, 0, + CTR_MSG_INSERT_SETATTR_WIND_FAILED, + "Failed to insert fsetxattr wind"); + } +out: + STACK_WIND (frame, ctr_fsetxattr_cbk, + FIRST_CHILD (this), FIRST_CHILD (this)->fops->fsetxattr, + fd, dict, flags, xdata); + return 0; +} /****************************mknod*******************************************/ @@ -1083,7 +1203,7 @@ ctr_mknod_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_MKNOD_UNWIND_FAILED, - "Failed inserting mknod unwind"); + "Failed to insert mknod unwind"); } out: @@ -1134,7 +1254,7 @@ ctr_mknod (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_MKNOD_WIND_FAILED, - "Failed inserting mknod wind"); + "Failed to insert mknod wind"); } out: @@ -1169,7 +1289,7 @@ ctr_create_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_CREATE_UNWIND_FAILED, - "Failed inserting create unwind"); + "Failed to insert create unwind"); } out: @@ -1232,7 +1352,7 @@ ctr_create (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_CREATE_WIND_FAILED, - "Failed inserting create wind"); + "Failed to insert create wind"); } out: STACK_WIND (frame, ctr_create_cbk, FIRST_CHILD (this), @@ -1264,7 +1384,7 @@ ctr_link_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_CREATE_UNWIND_FAILED, - "Failed inserting create unwind"); + "Failed to insert create unwind"); } out: @@ -1301,6 +1421,9 @@ ctr_link (call_frame_t *frame, xlator_t *this, /*Internal FOP*/ _inode_cx->is_internal_fop = is_internal_fop (frame, xdata); + /* Is a metadata fop */ + _inode_cx->is_metadata_fop = _gf_true; + /* If its a internal FOP and dht link file donot record*/ if (_inode_cx->is_internal_fop && dht_is_linkfile (&dummy_stat, xdata)) { @@ -1313,7 +1436,7 @@ ctr_link (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_LINK_WIND_FAILED, - "Failed inserting link wind"); + "Failed to insert link wind"); } out: @@ -1338,7 +1461,7 @@ int ctr_readv_cbk (call_frame_t *frame, void *cookie, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_CREATE_UNWIND_FAILED, - "Failed inserting create unwind"); + "Failed to insert create unwind"); } out: @@ -1369,7 +1492,7 @@ ctr_readv (call_frame_t *frame, xlator_t *this, if (ret) { gf_msg (this->name, GF_LOG_ERROR, 0, CTR_MSG_INSERT_READV_WIND_FAILED, - "Failed inserting readv wind"); + "Failed to insert readv wind"); } out: @@ -1692,6 +1815,10 @@ reconfigure (xlator_t *this, dict_t *options) GF_OPTION_RECONF ("record-counters", priv->ctr_record_counter, options, bool, out); + GF_OPTION_RECONF ("ctr-record-metadata-heat", + priv->ctr_record_metadata_heat, options, + bool, out); + GF_OPTION_RECONF ("ctr_link_consistency", priv->ctr_link_consistency, options, bool, out); @@ -1874,23 +2001,26 @@ fini (xlator_t *this) struct xlator_fops fops = { /*lookup*/ - .lookup = ctr_lookup, + .lookup = ctr_lookup, /*write fops */ - .mknod = ctr_mknod, - .create = ctr_create, - .truncate = ctr_truncate, - .ftruncate = ctr_ftruncate, - .setxattr = ctr_setxattr, - .removexattr = ctr_removexattr, - .unlink = ctr_unlink, - .link = ctr_link, - .rename = ctr_rename, - .writev = ctr_writev, - .setattr = ctr_setattr, + .mknod = ctr_mknod, + .create = ctr_create, + .truncate = ctr_truncate, + .ftruncate = ctr_ftruncate, + .setxattr = ctr_setxattr, + .fsetxattr = ctr_fsetxattr, + .removexattr = ctr_removexattr, + .fremovexattr = ctr_fremovexattr, + .unlink = ctr_unlink, + .link = ctr_link, + .rename = ctr_rename, + .writev = ctr_writev, + .setattr = ctr_setattr, + .fsetattr = ctr_fsetattr, /*read fops*/ - .readv = ctr_readv, + .readv = ctr_readv, /* IPC call*/ - .ipc = ctr_ipc + .ipc = ctr_ipc }; struct xlator_cbks cbks = { @@ -1919,6 +2049,11 @@ struct volume_options options[] = { .value = {"on", "off"}, .default_value = "off" }, + { .key = {"ctr-record-metadata-heat"}, + .type = GF_OPTION_TYPE_BOOL, + .value = {"on", "off"}, + .default_value = "off" + }, { .key = {"ctr_link_consistency"}, .type = GF_OPTION_TYPE_BOOL, .value = {"on", "off"}, diff --git a/xlators/features/changetimerecorder/src/ctr-helper.c b/xlators/features/changetimerecorder/src/ctr-helper.c index 047abd9f032..ab918eac825 100644 --- a/xlators/features/changetimerecorder/src/ctr-helper.c +++ b/xlators/features/changetimerecorder/src/ctr-helper.c @@ -272,6 +272,11 @@ int extract_ctr_options (xlator_t *this, gf_ctr_private_t *_priv) { GF_OPTION_INIT ("record-counters", _priv->ctr_record_counter, bool, out); + /* Extract flag for record metadata heat */ + GF_OPTION_INIT ("ctr-record-metadata-heat", + _priv->ctr_record_metadata_heat, bool, + out); + /*Extract flag for link consistency*/ GF_OPTION_INIT ("ctr_link_consistency", _priv->ctr_link_consistency, bool, out); diff --git a/xlators/features/changetimerecorder/src/ctr-helper.h b/xlators/features/changetimerecorder/src/ctr-helper.h index 5599bdd428b..161fff2d774 100644 --- a/xlators/features/changetimerecorder/src/ctr-helper.h +++ b/xlators/features/changetimerecorder/src/ctr-helper.h @@ -45,6 +45,7 @@ typedef struct gf_ctr_private { gf_boolean_t ctr_record_wind; gf_boolean_t ctr_record_unwind; gf_boolean_t ctr_record_counter; + gf_boolean_t ctr_record_metadata_heat; gf_boolean_t ctr_link_consistency; gfdb_db_type_t gfdb_db_type; gfdb_sync_type_t gfdb_sync_type; @@ -78,7 +79,7 @@ typedef struct gf_ctr_local { gfdb_db_record_t gfdb_db_record; ia_type_t ia_inode_type; gf_boolean_t is_internal_fop; - gf_client_pid_t client_pid; + gf_client_pid_t client_pid; } gf_ctr_local_t; /* * Easy access of gfdb_db_record of ctr_local @@ -171,6 +172,8 @@ typedef struct gf_ctr_inode_context { gfdb_fop_type_t fop_type; gfdb_fop_path_t fop_path; gf_boolean_t is_internal_fop; + /* Indicating metadata fops */ + gf_boolean_t is_metadata_fop; } gf_ctr_inode_context_t; @@ -338,6 +341,19 @@ do {\ goto label;\ } while (0) +/* + * IS CTR record metadata heat is disabled then goto to label + * */ + #define CTR_RECORD_METADATA_HEAT_IS_DISABLED_THEN_GOTO(this, label)\ + do {\ + gf_ctr_private_t *_priv = NULL;\ + GF_ASSERT (this);\ + GF_ASSERT (this->private);\ + _priv = this->private;\ + if (!_priv->ctr_record_metadata_heat)\ + goto label;\ + } while (0) + int fill_db_record_for_unwind (xlator_t *this, gf_ctr_local_t *ctr_local, @@ -390,16 +406,41 @@ ctr_insert_wind (call_frame_t *frame, ctr_local->is_internal_fop = ctr_inode_cx->is_internal_fop; /* Decide whether to record counters or not */ - CTR_DB_REC(ctr_local).do_record_counters = - _priv->ctr_record_counter && - !(ctr_local->is_internal_fop); + CTR_DB_REC(ctr_local).do_record_counters = _gf_false; + /* If record counter is enabled */ + if (_priv->ctr_record_counter) { + /* If not a internal fop */ + if (!(ctr_local->is_internal_fop)) { + /* If its a metadata fop AND + * record metadata heat + * OR + * its NOT a metadata fop */ + if ((ctr_inode_cx->is_metadata_fop + && _priv->ctr_record_metadata_heat) + || + (!ctr_inode_cx->is_metadata_fop)) { + CTR_DB_REC(ctr_local).do_record_counters + = _gf_true; + } + } + } /* Decide whether to record times or not * For non internal FOPS record times as usual*/ + CTR_DB_REC(ctr_local).do_record_times = _gf_false; if (!ctr_local->is_internal_fop) { - CTR_DB_REC(ctr_local).do_record_times = - (_priv->ctr_record_wind - || _priv->ctr_record_unwind); + /* If its a metadata fop AND + * record metadata heat + * OR + * its NOT a metadata fop */ + if ((ctr_inode_cx->is_metadata_fop && + _priv->ctr_record_metadata_heat) + || + (!ctr_inode_cx->is_metadata_fop)) { + CTR_DB_REC(ctr_local).do_record_times = + (_priv->ctr_record_wind + || _priv->ctr_record_unwind); + } } /* when its a internal FOPS*/ else { diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c index ccf4b380ffb..e3110b060fe 100644 --- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c +++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c @@ -2116,6 +2116,22 @@ struct volopt_map_entry glusterd_volopt_map[] = { "of writes (or reads) to a given file are needed " "before triggering migration." }, + { .key = "features.ctr-record-metadata-heat", + .voltype = "features/changetimerecorder", + .value = "off", + .option = "ctr-record-metadata-heat", + .op_version = GD_OP_VERSION_3_7_0, + /* Purposefully commenting the description so that this option remains + * hidden from the users as this is more of a developer option as of + * now. + * .description = "Its a Change Time Recorder Xlator option to " + * "enable recording write heat on metadata of the file. " + "The default is disabled. " + "Metadata is inode atttributes like atime, mtime," + " permissions etc and " + "extended attributes of a file ." + * */ + }, { .key = "features.ctr_link_consistency", .voltype = "features/changetimerecorder", .value = "off", -- cgit