local old_fid="$1"
local path="$2"
- echo -e "$old_fid $path" >> "$MIGRATED_SET"
+ echo "$old_fid $path" >> "$MIGRATED_SET"
}
path_in_set() {
cat -- <<USAGE 1>&2
usage: lfs_migrate [--dry-run|-n] [--help|-h] [--no-rsync|--rsync] [--quiet|-q]
[--auto-stripe|-A [-C <cap>]
- [--min-free|-M <min_free>] [--max-free|-X <max_free>]]
- [--stripe-count|-c <stripe_count>]
+ [--min-free|-M <min_free>] [--max-free|-X <max_free>]]
+ [--pool|-p <pool>] [--stripe-count|-c <stripe_count>]
[--stripe-size|-S <stripe_size>]
[-D] [-h] [-n] [-S]
[--restripe|-R] [--skip|-s] [--verbose|-v] [--yes|-y] [-0]
use in the migration
--no-rsync do not fall back to rsync mode even if lfs migrate fails
-n only print the names of files to be migrated
+ -p <pool> use the specified OST pool for the destination file
-q run quietly (don't print filenames or status)
--rsync force rsync mode instead of using lfs migrate
-R restripe file using default directory striping
OPT_NO_DIRECT=false
OPT_NULL=false
OPT_PASSTHROUGH=()
+OPT_POOL=""
OPT_RESTRIPE=false
OPT_YES=false
LFS_OPT_DIRECTIO=""
case "$arg" in
-h|--help) usage;;
-l|--link) ;; # maintained backward compatibility for now
- -n|--dry-run) OPT_DRYRUN=true; OPT_YES=true
+ -n) OPT_DRYRUN=true; OPT_YES=true
echo "$PROG: -n deprecated, use --dry-run or --non-block" 1>&2;;
+ --dry-run) OPT_DRYRUN=true; OPT_YES=true;;
+ -p|--pool) OPT_POOL="$arg $2"; OPT_LAYOUT+="$OPT_POOL "; shift;;
-q|--quiet) ECHO=:;;
-R|--restripe) OPT_RESTRIPE=true;;
-s|--skip) OPT_CHECK=false;;
- -v|--verbose) OPT_DEBUG=true; ECHO=echo; OPT_PASSTHROUGH+=("$arg");;
+ -v|--verbose) OPT_DEBUG=true; ECHO=echo;;
-y|--yes) OPT_YES=true;;
-0) OPT_NULL=true;;
-b|--block|--non-block|--non-direct|--no-verify)
--no-rsync) OPT_NO_RSYNC=true;;
--copy|--yaml|--file)
# these options have files as arguments, pass both through
- OPT_LAYOUT+="$arg $2"; shift;;
+ OPT_LAYOUT+="$arg $2 "; shift;;
--auto-stripe|-A) OPT_AUTOSTRIPE=true;;
-C) OPT_CAP="$2"; shift;;
-D) LFS_OPT_DIRECTIO="-D";;
ost_min_kb=$avail
fi
fi
- done < <($LFS df $OLDNAME | awk '/OST/ { print $4 }')
- # Once this script supports pools, the lfs df command above
- # should also include the -p <pool> option to restrict the
- # listed OSTs to the correct pool.
+ done < <($LFS df $OPT_POOL $OLDNAME | awk '/OST/ { print $4 }')
if [ $ost_max_count -eq 0 ]; then
- echo "no OSTs with sufficient available space" >&2
+ # no OSTs with enough space, stripe over all of them
+ echo "-1" "0"
return
fi
- if [ "$ost_min_kb" -eq $((1 << 62)) ]; then
+ if (( ost_min_kb == (1 << 62) )); then
echo "warning: unable to determine minimum OST size, " \
"object size not capped" >&2
- obj_max_kb=0
- echo "$stripe_count" "$obj_max_kb"
+ echo "$stripe_count" "0"
return
fi
return
fi
- # If disk usage would exceed the cap, increase the number of stripes
- [ $filekb -gt $((stripe_count * $obj_max_kb)) ] &&
- stripe_count=$((filekb / $obj_max_kb))
+ # If disk usage would exceed the cap, increase the number of stripes.
+ # Round up to the nearest MB to ensure file will fit.
+ (( filekb > stripe_count * obj_max_kb )) &&
+ stripe_count=$(((filekb + obj_max_kb - 1024) / obj_max_kb))
# Limit the count to the number of eligible OSTs
if [ "$stripe_count" -gt $ost_max_count ]; then
}
lfs_migrate() {
+ local last_dev
+ local mntpoint
+
while IFS='' read -d '' OLDNAME; do
local hlinks=()
local stripe_size="$OPT_STRIPE_SIZE"
local stripe_count="$OPT_STRIPE_COUNT"
local parent_count=""
local parent_size=""
- local stripe_pool
+ local stripe_pool="${OPT_POOL#-p }"
local mirror_count
local layout
+ local fid
$ECHO -n "$OLDNAME: "
- # avoid duplicate stat if possible
- local nlink_type=($(LANG=C stat -c "%h %F %s" "$OLDNAME" \
+ # avoid duplicate stat call by fetching all attrs at once
+ local nlink_idx_link=0 # %h is the hard link count
+ local nlink_idx_type=1 # %F is "regular file", ignore others
+ local nlink_idx_file=2 # "file" is here
+ local nlink_idx_size=3 # %s is file size in bytes
+ local nlink_idx_dev=4 # %D is the underlying device number
+ # nlink_type=(1 regular file 1234 0x810)
+ local nlink_type=($(LANG=C stat -c "%h %F %s %D" "$OLDNAME" \
2> /dev/null))
# skip non-regular files, since they don't have any objects
# and there is no point in trying to migrate them.
- if [ "${nlink_type[1]}" != "regular" ]; then
- echo -e "$OLDNAME: not a regular file, skipped" 1>&2
+ if [ "${nlink_type[$nlink_idx_type]}" != "regular" ]; then
+ echo -e "\r$OLDNAME: not a regular file, skipped" 1>&2
continue
fi
# working out write perms is hard, let the shell do it
if [ ! -w "$OLDNAME" ]; then
- echo -e "$OLDNAME: no write permission, skipped" 1>&2
+ echo -e "\r$OLDNAME: no write permission, skipped" 1>&2
continue
fi
# also absolute so that the names can be compared
local oldname_absolute=$(readlink -f "$OLDNAME")
if [ -z "$oldname_absolute" ]; then
- echo -e "$OLDNAME: cannot resolve full path, skipped" 1>&2
+ echo -e "\r$OLDNAME: cannot resolve full path, skipped" 1>&2
continue
fi
OLDNAME=$oldname_absolute
- # In the future, the path2fid and fid2path calls below
- # should be replaced with a single call to
- # "lfs path2links" once that command is available. The logic
- # for detecting unlisted hard links could then be removed.
- local fid=$($LFS path2fid "$OLDNAME" 2> /dev/null)
- if [ $? -ne 0 ]; then
- echo -n "\r\e[K$OLDNAME: cannot determine FID; skipping; "
- echo "is this a Lustre file system?"
- echo -e "$OLDNAME: cannot determine FID; skipping; " 1>&2
- echo "is this a Lustre file system?" 1>&2
- continue
- fi
+ if [[ ${nlink_type[$nlink_idx_link]} -gt 1 ]] ||
+ $RSYNC_WITH_HLINKS; then
+ fid=$($LFS path2fid "$OLDNAME" 2> /dev/null)
+ if [ $? -ne 0 ]; then
+ echo -e "\r$OLDNAME: cannot get FID, skipping; is this a Lustre file system?" 1>&2
+ continue
+ fi
- if [[ ${nlink_type[0]} -gt 1 ]] || $RSYNC_WITH_HLINKS; then
# don't migrate a hard link if it was already migrated
if path_in_set "$OLDNAME"; then
- $ECHO "\r\e[Kalready migrated via another hard link"
+ $ECHO "already migrated via another hard link"
continue
fi
# being migrated.
local migrated=$(old_fid_in_set "$fid")
if [ -n "$migrated" ]; then
- $ECHO -e "$OLDNAME: already migrated via another hard link"
- if $OPT_RSYNC; then
- # Only the rsync case has to relink.
- # The lfs migrate case preserves the
- # inode so the links are already
- # correct.
- [ "$migrated" != "$OLDNAME" ] &&
- ln -f "$migrated" "$OLDNAME"
- fi
+ $ECHO "already migrated via another hard link"
+ # Only the rsync case has to relink. The
+ # "lfs migrate" case keeps the same inode so
+ # all of the links are already correct.
+ $OPT_RSYNC && [ "$migrated" != "$OLDNAME" ] &&
+ ln -f "$migrated" "$OLDNAME"
+
add_to_set "$fid" "$OLDNAME"
continue;
fi
if $OPT_RESTRIPE; then
UNLINK=""
else
- # if rsync copies Lustre xattrs properly in the future
+ # If rsync copies Lustre xattrs properly in the future
# (i.e. before the file data, so that it preserves
- # striping) then we don't need to do this getstripe
- # stuff.
+ # striping) then we don't need this getstripe stuff.
UNLINK="-u"
- stripe_pool=$($LFS getstripe -p "$OLDNAME" 2> /dev/null)
- mirror_count=$($LFS getstripe -N "$OLDFILE" 2> /dev/null)
+ [ -n "$OPT_POOL" ] ||
+ stripe_pool=$($LFS getstripe -p "$OLDNAME" 2>/dev/null)
+ mirror_count=$($LFS getstripe -N "$OLDFILE" 2>/dev/null)
if $OPT_AUTOSTRIPE; then
- local filekb=$((${nlink_type[3]} / 1024))
+ local filekb=$((${nlink_type[$nlink_idx_size]} /
+ 1024))
+
read stripe_count OBJ_MAX_KB < <(calc_stripe \
"$OLDNAME" "$filekb" "$OBJ_MAX_KB")
[ -z "$stripe_count" ] && exit 1
[ $stripe_count -lt 1 ] && stripe_count=1
else
- [ "$OPT_STRIPE_COUNT" ] && stripe_count=$OPT_STRIPE_COUNT ||
+ [ -n "$stripe_count" ] ||
stripe_count=$($LFS getstripe -c "$OLDNAME" \
2> /dev/null)
fi
- [ -z "$stripe_size" ] &&
+ [ -n "$stripe_size" ] ||
stripe_size=$($LFS getstripe -S "$OLDNAME" 2> /dev/null)
[ -z "$stripe_count" -o -z "$stripe_size" ] && UNLINK=""
/dev/null)
fi
- $ECHO -n "stripe" \
- "count=${stripe_count:-$parent_count}," \
- "size=${stripe_size:-$parent_size}," \
- "pool=${stripe_pool}," \
- "mirror_count=${mirror_count}"
+ $ECHO -n "stripe_count=${stripe_count:-$parent_count},stripe_size=${stripe_size:-$parent_size}"
+ [ -n "$stripe_pool" ] &&
+ $ECHO -n ",pool=${stripe_pool}"
+ [ -n "$mirror_count" ] &&
+ $ECHO -n ",mirror_count=${mirror_count}"
+ $ECHO -n " "
fi
if $OPT_DRYRUN; then
- $ECHO "dry run, skipped"
+ $ECHO " dry run, skipped"
continue
fi
- [ -n "$stripe_count" ] && stripe_count="-c $stripe_count"
- [ -n "$stripe_size" ] && stripe_size="-S $stripe_size"
- [ -n "$stripe_pool" ] && stripe_pool="-p $stripe_pool"
- [ -n "$mirror_count" ] && mirror_count="-N $mirror_count"
- layout="$stripe_count $stripe_size $stripe_pool $mirror_count \
- $OPT_LAYOUT"
+ layout="${OPT_PASSTHROUGH[@]} "
+ [ -n "$stripe_count" ] && layout+="-c $stripe_count "
+ [ -n "$stripe_size" ] && layout+="-S $stripe_size "
+ [ -z "$OPT_POOL" -a -n "$stripe_pool" ] &&
+ layout+="-p $stripe_pool "
+ [ -n "$mirror_count" ] && layout+="-N $mirror_count "
+ layout+="$OPT_LAYOUT"
# detect other hard links and store them on a global
# list so we don't re-migrate them
- local mntpoint=$(df -P "$OLDNAME" |
- awk 'NR==2 { print $NF; exit }')
- if [ -z "$mntpoint" ]; then
- echo -e "$OLDNAME: cannot determine mount point; skipped" 1>&2
- continue
- fi
- hlinks=$($LFS fid2path "$mntpoint" "$fid" 2> /dev/null)
- if [ $? -ne 0 ]; then
- echo -e "$OLDNAME: cannot determine hard link paths, skipped" 1>&2
- continue
+ if [[ ${nlink_type[$nlink_idx_link]} -gt 1 ]]; then
+ [ "${nlink_type[$nlink_idx_dev]}" == "$last_dev" ] ||
+ mntpoint=$(df -P "$OLDNAME" |
+ awk 'NR==2 { print $NF }')
+ if [ -z "$mntpoint" ]; then
+ echo -e "\r$OLDNAME: cannot determine mount point; skipped" 1>&2
+ continue
+ fi
+ hlinks=$($LFS fid2path "$mntpoint" "$fid" 2> /dev/null)
+ if $OPT_RSYNC && [ $? -ne 0 ]; then
+ echo -e "\r$OLDNAME: cannot determine hard link paths, skipped" 1>&2
+ continue
+ fi
+ hlinks+=("$OLDNAME")
+ else
+ hlinks=
fi
- hlinks+=("$OLDNAME")
# first try to migrate via Lustre tools, then fall back to rsync
if ! $OPT_RSYNC; then
- if $LFS migrate "${OPT_PASSTHROUGH[@]}" $layout \
- "$OLDNAME"; then
- $ECHO "done migrate"
+ if $LFS migrate $layout "$OLDNAME"; then
+ $ECHO "done"
+ # no-op if hlinks empty for 1-link files
for link in ${hlinks[*]}; do
add_to_set "$fid" "$link"
done
continue
elif $OPT_NO_RSYNC; then
- echo -e "$OLDNAME: refusing to fall back to rsync, skipped" 1>&2
+ echo -e "\r$OLDNAME: refusing to fall back to rsync, skipped" 1>&2
continue
else
$ECHO -n "falling back to rsync: "
fi
fi
- NEWNAME=$(mktemp $UNLINK "$OLDNAME-lfs_migrate.tmp.XXXXXX")
+ local olddir=$(dirname $OLDNAME)
+ local oldfile=$(basename $OLDNAME)
+ NEWNAME=$(mktemp $UNLINK "$olddir/.$oldfile.XXXXXX")
if [ $? -ne 0 -o -z "$NEWNAME" ]; then
- echo -e "$OLDNAME: cannot make temp file, skipped" 1>&2
+ echo -e "\r$OLDNAME: cannot make temp file, skipped" 1>&2
continue
fi
if [ "$UNLINK" ]; then
- if ! $LFS setstripe "${OPT_PASSTHROUGH[@]}" $layout \
- "$NEWNAME"; then
- echo -e "\r\e[K$NEWNAME: setstripe failed, exiting" 1>&2
+ if ! $LFS setstripe $layout "$NEWNAME"; then
+ echo -e "\r$NEWNAME: setstripe failed, exiting" 1>&2
exit 2
fi
fi
# we use --inplace, since we created our own temp file already
if ! $RSYNC -a --inplace $RSYNC_OPTS "$OLDNAME" "$NEWNAME";then
- echo -e "$OLDNAME: copy error, exiting" 1>&2
+ echo -e "\r$OLDNAME: copy error, exiting" 1>&2
exit 4
fi
if $OPT_CHECK && ! cmp -s "$OLDNAME" "$NEWNAME"; then
- echo -e "$NEWNAME: compare failed, exiting" 1>&2
+ echo -e "\r$NEWNAME: compare failed, exiting" 1>&2
exit 8
fi
if ! mv "$NEWNAME" "$OLDNAME"; then
- echo -e "$OLDNAME: rename error, exiting" 1>&2
+ echo -e "\r$OLDNAME: rename error, exiting" 1>&2
exit 12
fi
- $ECHO "done migrate via rsync"
+ $ECHO "done rsync"
+ # no-op if hlinks empty for 1-link files
for link in ${hlinks[*]}; do
if [ "$link" != "$OLDNAME" ]; then
ln -f "$OLDNAME" "$link"
# of 1 (all other links will point to the new inode).
# This flag indicates that even paths with a link count of
# 1 are potentially part of a link set.
- [ ${#hlinks[*]} -gt 1 ] && RSYNC_WITH_HLINKS=true
+ (( ${#hlinks[*]} == 1 )) || RSYNC_WITH_HLINKS=true
done
}