ceph/src/vstart.sh

   1 #!/bin/bash
   2
   3 # abort on failure
   4 set -e
   5
   6 if [ -n "$VSTART_DEST" ]; then
   7   SRC_PATH=`dirname $0`
   8   SRC_PATH=`(cd $SRC_PATH; pwd)`
   9
  10   CEPH_DIR=$SRC_PATH
  11   CEPH_BIN=${PWD}/bin
  12   CEPH_LIB=${PWD}/lib
  13
  14   CEPH_CONF_PATH=$VSTART_DEST
  15   CEPH_DEV_DIR=$VSTART_DEST/dev
  16   CEPH_OUT_DIR=$VSTART_DEST/out
  17 fi
  18
  19 # for running out of the CMake build directory
  20 if [ -e CMakeCache.txt ]; then
  21   # Out of tree build, learn source location from CMakeCache.txt
  22   CEPH_ROOT=`grep ceph_SOURCE_DIR CMakeCache.txt | cut -d "=" -f 2`
  23   CEPH_BUILD_DIR=`pwd`
  24   [ -z "$MGR_PYTHON_PATH" ] && MGR_PYTHON_PATH=$CEPH_ROOT/src/pybind/mgr
  25 fi
  26
  27 # use CEPH_BUILD_ROOT to vstart from a 'make install'
  28 if [ -n "$CEPH_BUILD_ROOT" ]; then
  29         [ -z "$CEPH_BIN" ] && CEPH_BIN=$CEPH_BUILD_ROOT/bin
  30         [ -z "$CEPH_LIB" ] && CEPH_LIB=$CEPH_BUILD_ROOT/lib
  31         [ -z "$EC_PATH" ] && EC_PATH=$CEPH_LIB/erasure-code
  32         [ -z "$OBJCLASS_PATH" ] && OBJCLASS_PATH=$CEPH_LIB/rados-classes
  33 elif [ -n "$CEPH_ROOT" ]; then
  34         [ -z "$PYBIND" ] && PYBIND=$CEPH_ROOT/src/pybind
  35         [ -z "$CEPH_BIN" ] && CEPH_BIN=$CEPH_BUILD_DIR/bin
  36         [ -z "$CEPH_ADM" ] && CEPH_ADM=$CEPH_BIN/ceph
  37         [ -z "$INIT_CEPH" ] && INIT_CEPH=$CEPH_BUILD_DIR/bin/init-ceph
  38         [ -z "$CEPH_LIB" ] && CEPH_LIB=$CEPH_BUILD_DIR/lib
  39         [ -z "$OBJCLASS_PATH" ] && OBJCLASS_PATH=$CEPH_LIB
  40         [ -z "$EC_PATH" ] && EC_PATH=$CEPH_LIB
  41 fi
  42
  43 if [ -z "${CEPH_VSTART_WRAPPER}" ]; then
  44     PATH=$(pwd):$PATH
  45 fi
  46
  47 [ -z "$PYBIND" ] && PYBIND=./pybind
  48
  49 export PYTHONPATH=$PYBIND:$CEPH_LIB/cython_modules/lib.2:$PYTHONPATH
  50 export LD_LIBRARY_PATH=$CEPH_LIB:$LD_LIBRARY_PATH
  51 export DYLD_LIBRARY_PATH=$CEPH_LIB:$DYLD_LIBRARY_PATH
  52 # Suppress logging for regular use that indicated that we are using a
  53 # development version. vstart.sh is only used during testing and
  54 # development
  55 export CEPH_DEV=1
  56
  57 [ -z "$CEPH_NUM_MON" ] && CEPH_NUM_MON="$MON"
  58 [ -z "$CEPH_NUM_OSD" ] && CEPH_NUM_OSD="$OSD"
  59 [ -z "$CEPH_NUM_MDS" ] && CEPH_NUM_MDS="$MDS"
  60 [ -z "$CEPH_NUM_MGR" ] && CEPH_NUM_MGR="$MGR"
  61 [ -z "$CEPH_NUM_FS"  ] && CEPH_NUM_FS="$FS"
  62 [ -z "$CEPH_NUM_RGW" ] && CEPH_NUM_RGW="$RGW"
  63
  64 # if none of the CEPH_NUM_* number is specified, kill the existing
  65 # cluster.
  66 if [ -z "$CEPH_NUM_MON" -a \
  67      -z "$CEPH_NUM_OSD" -a \
  68      -z "$CEPH_NUM_MDS" -a \
  69      -z "$CEPH_NUM_MGR" ]; then
  70     kill_all=1
  71 else
  72     kill_all=0
  73 fi
  74
  75 [ -z "$CEPH_NUM_MON" ] && CEPH_NUM_MON=3
  76 [ -z "$CEPH_NUM_OSD" ] && CEPH_NUM_OSD=3
  77 [ -z "$CEPH_NUM_MDS" ] && CEPH_NUM_MDS=3
  78 [ -z "$CEPH_NUM_MGR" ] && CEPH_NUM_MGR=1
  79 [ -z "$CEPH_NUM_FS"  ] && CEPH_NUM_FS=1
  80 [ -z "$CEPH_MAX_MDS" ] && CEPH_MAX_MDS=1
  81 [ -z "$CEPH_NUM_RGW" ] && CEPH_NUM_RGW=0
  82
  83 [ -z "$CEPH_DIR" ] && CEPH_DIR="$PWD"
  84 [ -z "$CEPH_DEV_DIR" ] && CEPH_DEV_DIR="$CEPH_DIR/dev"
  85 [ -z "$CEPH_OUT_DIR" ] && CEPH_OUT_DIR="$CEPH_DIR/out"
  86 [ -z "$CEPH_RGW_PORT" ] && CEPH_RGW_PORT=8000
  87 [ -z "$CEPH_CONF_PATH" ] && CEPH_CONF_PATH=$CEPH_DIR
  88
  89 if [ $CEPH_NUM_OSD -gt 3 ]; then
  90     OSD_POOL_DEFAULT_SIZE=3
  91 else
  92     OSD_POOL_DEFAULT_SIZE=$CEPH_NUM_OSD
  93 fi
  94
  95 extra_conf=""
  96 new=0
  97 standby=0
  98 debug=0
  99 ip=""
 100 nodaemon=0
 101 smallmds=0
 102 short=0
 103 ec=0
 104 hitset=""
 105 overwrite_conf=1
 106 cephx=1 #turn cephx on by default
 107 cache=""
 108 memstore=0
 109 bluestore=0
 110 rgw_frontend="civetweb"
 111 rgw_compression=""
 112 lockdep=${LOCKDEP:-1}
 113
 114 filestore_path=
 115
 116 VSTART_SEC="client.vstart.sh"
 117
 118 MON_ADDR=""
 119 DASH_URLS=""
 120 RESTFUL_URLS=""
 121
 122 conf_fn="$CEPH_CONF_PATH/ceph.conf"
 123 keyring_fn="$CEPH_CONF_PATH/keyring"
 124 osdmap_fn="/tmp/ceph_osdmap.$$"
 125 monmap_fn="/tmp/ceph_monmap.$$"
 126
 127 usage="usage: $0 [option]... \nex: $0 -n -d --mon_num 3 --osd_num 3 --mds_num 1 --rgw_num 1\n"
 128 usage=$usage"options:\n"
 129 usage=$usage"\t-d, --debug\n"
 130 usage=$usage"\t-s, --standby_mds: Generate standby-replay MDS for each active\n"
 131 usage=$usage"\t-l, --localhost: use localhost instead of hostname\n"
 132 usage=$usage"\t-i <ip>: bind to specific ip\n"
 133 usage=$usage"\t-n, --new\n"
 134 usage=$usage"\t-N, --not-new: reuse existing cluster config (default)\n"
 135 usage=$usage"\t--valgrind[_{osd,mds,mon,rgw}] 'toolname args...'\n"
 136 usage=$usage"\t--nodaemon: use ceph-run as wrapper for mon/osd/mds\n"
 137 usage=$usage"\t--smallmds: limit mds cache size\n"
 138 usage=$usage"\t-m ip:port\t\tspecify monitor address\n"
 139 usage=$usage"\t-k keep old configuration files\n"
 140 usage=$usage"\t-x enable cephx (on by default)\n"
 141 usage=$usage"\t-X disable cephx\n"
 142 usage=$usage"\t--hitset <pool> <hit_set_type>: enable hitset tracking\n"
 143 usage=$usage"\t-e : create an erasure pool\n";
 144 usage=$usage"\t-o config\t\t add extra config parameters to all sections\n"
 145 usage=$usage"\t--mon_num specify ceph monitor count\n"
 146 usage=$usage"\t--osd_num specify ceph osd count\n"
 147 usage=$usage"\t--mds_num specify ceph mds count\n"
 148 usage=$usage"\t--rgw_num specify ceph rgw count\n"
 149 usage=$usage"\t--mgr_num specify ceph mgr count\n"
 150 usage=$usage"\t--rgw_port specify ceph rgw http listen port\n"
 151 usage=$usage"\t--rgw_frontend specify the rgw frontend configuration\n"
 152 usage=$usage"\t--rgw_compression specify the rgw compression plugin\n"
 153 usage=$usage"\t-b, --bluestore use bluestore as the osd objectstore backend\n"
 154 usage=$usage"\t--memstore use memstore as the osd objectstore backend\n"
 155 usage=$usage"\t--cache <pool>: enable cache tiering on pool\n"
 156 usage=$usage"\t--short: short object names only; necessary for ext4 dev\n"
 157 usage=$usage"\t--nolockdep disable lockdep\n"
 158 usage=$usage"\t--multimds <count> allow multimds with maximum active count\n"
 159
 160 usage_exit() {
 161         printf "$usage"
 162         exit
 163 }
 164
 165 while [ $# -ge 1 ]; do
 166 case $1 in
 167     -d | --debug )
 168             debug=1
 169             ;;
 170     -s | --standby_mds)
 171             standby=1
 172             ;;
 173     -l | --localhost )
 174             ip="127.0.0.1"
 175             ;;
 176     -i )
 177             [ -z "$2" ] && usage_exit
 178             ip="$2"
 179             shift
 180             ;;
 181     -e )
 182             ec=1
 183             ;;
 184     --new | -n )
 185             new=1
 186             ;;
 187     --not-new | -N )
 188         new=0
 189         ;;
 190     --short )
 191             short=1
 192             ;;
 193     --valgrind )
 194             [ -z "$2" ] && usage_exit
 195             valgrind=$2
 196             shift
 197             ;;
 198     --valgrind_args )
 199         valgrind_args="$2"
 200         shift
 201         ;;
 202     --valgrind_mds )
 203             [ -z "$2" ] && usage_exit
 204             valgrind_mds=$2
 205             shift
 206             ;;
 207     --valgrind_osd )
 208             [ -z "$2" ] && usage_exit
 209             valgrind_osd=$2
 210             shift
 211             ;;
 212     --valgrind_mon )
 213             [ -z "$2" ] && usage_exit
 214             valgrind_mon=$2
 215             shift
 216             ;;
 217     --valgrind_mgr )
 218             [ -z "$2" ] && usage_exit
 219             valgrind_mgr=$2
 220             shift
 221             ;;
 222     --valgrind_rgw )
 223             [ -z "$2" ] && usage_exit
 224             valgrind_rgw=$2
 225             shift
 226             ;;
 227     --nodaemon )
 228             nodaemon=1
 229             ;;
 230     --smallmds )
 231             smallmds=1
 232             ;;
 233     --mon_num )
 234             echo "mon_num:$2"
 235             CEPH_NUM_MON="$2"
 236             shift
 237             ;;
 238     --osd_num )
 239             CEPH_NUM_OSD=$2
 240             shift
 241             ;;
 242     --mds_num )
 243             CEPH_NUM_MDS=$2
 244             shift
 245             ;;
 246     --rgw_num )
 247             CEPH_NUM_RGW=$2
 248             shift
 249             ;;
 250     --mgr_num )
 251             CEPH_NUM_MGR=$2
 252             shift
 253             ;;
 254     --rgw_port )
 255             CEPH_RGW_PORT=$2
 256             shift
 257             ;;
 258     --rgw_frontend )
 259             rgw_frontend=$2
 260             shift
 261             ;;
 262     --rgw_compression )
 263             rgw_compression=$2
 264             shift
 265             ;;
 266     --filestore_path )
 267         filestore_path=$2
 268         shift
 269         ;;
 270     -m )
 271             [ -z "$2" ] && usage_exit
 272             MON_ADDR=$2
 273             shift
 274             ;;
 275     -x )
 276             cephx=1 # this is on be default, flag exists for historical consistency
 277             ;;
 278     -X )
 279             cephx=0
 280             ;;
 281     -k )
 282             if [ ! -r $conf_fn ]; then
 283                 echo "cannot use old configuration: $conf_fn not readable." >&2
 284                 exit
 285             fi
 286             overwrite_conf=0
 287             ;;
 288     --memstore )
 289             memstore=1
 290             ;;
 291     -b | --bluestore )
 292             bluestore=1
 293             ;;
 294     --hitset )
 295             hitset="$hitset $2 $3"
 296             shift
 297             shift
 298             ;;
 299     -o )
 300             extra_conf="$extra_conf     $2
 301 "
 302             shift
 303             ;;
 304     --cache )
 305             if [ -z "$cache" ]; then
 306                 cache="$2"
 307             else
 308                 cache="$cache $2"
 309             fi
 310             shift
 311             ;;
 312     --nolockdep )
 313             lockdep=0
 314             ;;
 315     --multimds)
 316         CEPH_MAX_MDS="$2"
 317         shift
 318         ;;
 319     * )
 320             usage_exit
 321 esac
 322 shift
 323 done
 324
 325 if [ $kill_all -eq 1 ]; then
 326     $SUDO $INIT_CEPH stop
 327 fi
 328
 329 if [ "$overwrite_conf" -eq 0 ]; then
 330     CEPH_ASOK_DIR=`dirname $($CEPH_BIN/ceph-conf --show-config-value admin_socket)`
 331     mkdir -p $CEPH_ASOK_DIR
 332     MON=`$CEPH_BIN/ceph-conf -c $conf_fn --name $VSTART_SEC num_mon 2>/dev/null` && \
 333         CEPH_NUM_MON="$MON"
 334     OSD=`$CEPH_BIN/ceph-conf -c $conf_fn --name $VSTART_SEC num_osd 2>/dev/null` && \
 335         CEPH_NUM_OSD="$OSD"
 336     MDS=`$CEPH_BIN/ceph-conf -c $conf_fn --name $VSTART_SEC num_mds 2>/dev/null` && \
 337         CEPH_NUM_MDS="$MDS"
 338     MGR=`$CEPH_BIN/ceph-conf -c $conf_fn --name $VSTART_SEC num_mgr 2>/dev/null` && \
 339         CEPH_NUM_MGR="$MGR"
 340     RGW=`$CEPH_BIN/ceph-conf -c $conf_fn --name $VSTART_SEC num_rgw 2>/dev/null` && \
 341         CEPH_NUM_RGW="$RGW"
 342 else
 343     if [ "$new" -ne 0 ]; then
 344         # only delete if -n
 345         asok_dir=`dirname $($CEPH_BIN/ceph-conf --show-config-value admin_socket)`
 346         if [ $asok_dir != /var/run/ceph ]; then
 347             [ -d $asok_dir ] && rm -f $asok_dir/* && rmdir $asok_dir
 348         fi
 349         if [ -z "$CEPH_ASOK_DIR" ]; then
 350             CEPH_ASOK_DIR=`mktemp -u -d "${TMPDIR:-/tmp}/ceph-asok.XXXXXX"`
 351         fi
 352         [ -e "$conf_fn" ] && rm -- "$conf_fn"
 353     else
 354         CEPH_ASOK_DIR=`dirname $($CEPH_BIN/ceph-conf --show-config-value admin_socket)`
 355         # -k is implied... (doesn't make sense otherwise)
 356         overwrite_conf=0
 357     fi
 358 fi
 359
 360 ARGS="-c $conf_fn"
 361
 362 prunb() {
 363     echo "$* &"
 364     "$@" &
 365 }
 366
 367 prun() {
 368     echo "$*"
 369     "$@"
 370 }
 371
 372 run() {
 373     type=$1
 374     shift
 375     eval "valg=\$valgrind_$type"
 376     [ -z "$valg" ] && valg="$valgrind"
 377
 378     if [ -n "$valg" ]; then
 379         prunb valgrind --tool="$valg" $valgrind_args "$@" -f
 380         sleep 1
 381     else
 382         if [ "$nodaemon" -eq 0 ]; then
 383             prun "$@"
 384         else
 385             prunb ./ceph-run "$@" -f
 386         fi
 387     fi
 388 }
 389
 390 wconf() {
 391         if [ "$overwrite_conf" -eq 1 ]; then
 392                 cat >> "$conf_fn"
 393         fi
 394 }
 395
 396 prepare_conf() {
 397     local DAEMONOPTS="
 398         log file = $CEPH_OUT_DIR/\$name.log
 399         admin socket = $CEPH_ASOK_DIR/\$name.asok
 400         chdir = \"\"
 401         pid file = $CEPH_OUT_DIR/\$name.pid
 402         heartbeat file = $CEPH_OUT_DIR/\$name.heartbeat
 403 "
 404
 405     wconf <<EOF
 406 ; generated by vstart.sh on `date`
 407 [$VSTART_SEC]
 408         num mon = $CEPH_NUM_MON
 409         num osd = $CEPH_NUM_OSD
 410         num mds = $CEPH_NUM_MDS
 411         num mgr = $CEPH_NUM_MGR
 412         num rgw = $CEPH_NUM_RGW
 413
 414 [global]
 415         fsid = $(uuidgen)
 416         osd pg bits = 3
 417         osd pgp bits = 5  ; (invalid, but ceph should cope!)
 418         osd pool default size = $OSD_POOL_DEFAULT_SIZE
 419         osd crush chooseleaf type = 0
 420         osd pool default min size = 1
 421         osd failsafe full ratio = .99
 422         mon osd nearfull ratio = .99
 423         mon osd backfillfull ratio = .99
 424         mon osd reporter subtree level = osd
 425         mon osd full ratio = .99
 426         mon data avail warn = 2
 427         mon data avail crit = 1
 428         erasure code dir = $EC_PATH
 429         plugin dir = $CEPH_LIB
 430         osd pool default erasure code profile = plugin=jerasure technique=reed_sol_van k=2 m=1 crush-failure-domain=osd
 431         rgw frontends = $rgw_frontend port=$CEPH_RGW_PORT
 432         ; needed for s3tests
 433         rgw crypt s3 kms encryption keys = testkey-1=YmluCmJvb3N0CmJvb3N0LWJ1aWxkCmNlcGguY29uZgo= testkey-2=aWIKTWFrZWZpbGUKbWFuCm91dApzcmMKVGVzdGluZwo=
 434         rgw crypt require ssl = false
 435         rgw lc debug interval = 10
 436         filestore fd cache size = 32
 437         run dir = $CEPH_OUT_DIR
 438         enable experimental unrecoverable data corrupting features = *
 439 EOF
 440         if [ "$lockdep" -eq 1 ] ; then
 441                 wconf <<EOF
 442         lockdep = true
 443 EOF
 444         fi
 445         if [ "$cephx" -eq 1 ] ; then
 446                 wconf <<EOF
 447         auth cluster required = cephx
 448         auth service required = cephx
 449         auth client required = cephx
 450 EOF
 451         else
 452                 wconf <<EOF
 453         auth cluster required = none
 454         auth service required = none
 455         auth client required = none
 456 EOF
 457         fi
 458         if [ "$short" -eq 1 ]; then
 459                 COSDSHORT="        osd max object name len = 460
 460         osd max object namespace len = 64"
 461         fi
 462         wconf <<EOF
 463 [client]
 464         keyring = $keyring_fn
 465         log file = $CEPH_OUT_DIR/\$name.\$pid.log
 466         admin socket = $CEPH_ASOK_DIR/\$name.\$pid.asok
 467
 468 [client.rgw]
 469
 470 [mds]
 471 $DAEMONOPTS
 472 $CMDSDEBUG
 473         mds debug frag = true
 474         mds debug auth pins = true
 475         mds debug subtrees = true
 476         mds data = $CEPH_DEV_DIR/mds.\$id
 477         mds root ino uid = `id -u`
 478         mds root ino gid = `id -g`
 479 $extra_conf
 480 [mgr]
 481         mgr data = $CEPH_DEV_DIR/mgr.\$id
 482         mgr module path = $MGR_PYTHON_PATH
 483         mon reweight min pgs per osd = 4
 484         mon pg warn min per osd = 3
 485 $DAEMONOPTS
 486 $CMGRDEBUG
 487 $extra_conf
 488 [osd]
 489 $DAEMONOPTS
 490         osd_check_max_object_name_len_on_startup = false
 491         osd data = $CEPH_DEV_DIR/osd\$id
 492         osd journal = $CEPH_DEV_DIR/osd\$id/journal
 493         osd journal size = 100
 494         osd class tmp = out
 495         osd class dir = $OBJCLASS_PATH
 496         osd class load list = *
 497         osd class default list = *
 498         osd scrub load threshold = 2000.0
 499         osd debug op order = true
 500         osd debug misdirected ops = true
 501         filestore wbthrottle xfs ios start flusher = 10
 502         filestore wbthrottle xfs ios hard limit = 20
 503         filestore wbthrottle xfs inodes hard limit = 30
 504         filestore wbthrottle btrfs ios start flusher = 10
 505         filestore wbthrottle btrfs ios hard limit = 20
 506         filestore wbthrottle btrfs inodes hard limit = 30
 507         osd copyfrom max chunk = 524288
 508         bluestore fsck on mount = true
 509         bluestore block create = true
 510         bluestore block db path = $CEPH_DEV_DIR/osd\$id/block.db.file
 511         bluestore block db size = 67108864
 512         bluestore block db create = true
 513         bluestore block wal path = $CEPH_DEV_DIR/osd\$id/block.wal.file
 514         bluestore block wal size = 1048576000
 515         bluestore block wal create = true
 516 $COSDDEBUG
 517 $COSDMEMSTORE
 518 $COSDSHORT
 519 $extra_conf
 520 [mon]
 521         mgr initial modules = restful status dashboard balancer
 522         mon pg warn min per osd = 3
 523         mon osd allow primary affinity = true
 524         mon reweight min pgs per osd = 4
 525         mon osd prime pg temp = true
 526         crushtool = $CEPH_BIN/crushtool
 527         mon allow pool delete = true
 528 $DAEMONOPTS
 529 $CMONDEBUG
 530 $extra_conf
 531         mon cluster log file = $CEPH_OUT_DIR/cluster.mon.\$id.log
 532 [global]
 533 $extra_conf
 534 EOF
 535 }
 536
 537 start_mon() {
 538     local MONS=""
 539     local count=0
 540     for f in a b c d e f g h i j k l m n o p q r s t u v w x y z
 541     do
 542         [ $count -eq $CEPH_NUM_MON ] && break;
 543         count=$(($count + 1))
 544         if [ -z "$MONS" ];
 545         then
 546                 MONS="$f"
 547         else
 548                 MONS="$MONS $f"
 549         fi
 550     done
 551
 552     if [ "$new" -eq 1 ]; then
 553                 if [ `echo $IP | grep '^127\\.'` ]
 554                 then
 555                         echo
 556                         echo "NOTE: hostname resolves to loopback; remote hosts will not be able to"
 557                         echo "  connect.  either adjust /etc/hosts, or edit this script to use your"
 558                         echo "  machine's real IP."
 559                         echo
 560                 fi
 561
 562                 prun $SUDO "$CEPH_BIN/ceph-authtool" --create-keyring --gen-key --name=mon. "$keyring_fn" --cap mon 'allow *'
 563                 prun $SUDO "$CEPH_BIN/ceph-authtool" --gen-key --name=client.admin --set-uid=0 \
 564                         --cap mon 'allow *' \
 565                         --cap osd 'allow *' \
 566                         --cap mds 'allow *' \
 567                         --cap mgr 'allow *' \
 568                         "$keyring_fn"
 569
 570                 prun $SUDO "$CEPH_BIN/ceph-authtool" --gen-key --name=client.rgw \
 571                     --cap mon 'allow rw' \
 572                     --cap osd 'allow rwx' \
 573                     --cap mgr 'allow rw' \
 574                     "$keyring_fn"
 575
 576                 # build a fresh fs monmap, mon fs
 577                 local str=""
 578                 local count=0
 579                 for f in $MONS
 580                 do
 581                         str="$str --add $f $IP:$(($CEPH_PORT+$count))"
 582                         wconf <<EOF
 583 [mon.$f]
 584         host = $HOSTNAME
 585         mon data = $CEPH_DEV_DIR/mon.$f
 586         mon addr = $IP:$(($CEPH_PORT+$count))
 587 EOF
 588                         count=$(($count + 1))
 589                 done
 590                 prun "$CEPH_BIN/monmaptool" --create --clobber $str --print "$monmap_fn"
 591
 592                 for f in $MONS
 593                 do
 594                         prun rm -rf -- "$CEPH_DEV_DIR/mon.$f"
 595                         prun mkdir -p "$CEPH_DEV_DIR/mon.$f"
 596                         prun "$CEPH_BIN/ceph-mon" --mkfs -c "$conf_fn" -i "$f" --monmap="$monmap_fn" --keyring="$keyring_fn"
 597                 done
 598
 599                 prun rm -- "$monmap_fn"
 600         fi
 601
 602         # start monitors
 603         for f in $MONS
 604         do
 605                 run 'mon' $CEPH_BIN/ceph-mon -i $f $ARGS $CMON_ARGS
 606         done
 607 }
 608
 609 start_osd() {
 610     for osd in `seq 0 $((CEPH_NUM_OSD-1))`
 611     do
 612             if [ "$new" -eq 1 ]; then
 613                     wconf <<EOF
 614 [osd.$osd]
 615         host = $HOSTNAME
 616 EOF
 617
 618             rm -rf $CEPH_DEV_DIR/osd$osd || true
 619             if command -v btrfs > /dev/null; then
 620                 for f in $CEPH_DEV_DIR/osd$osd/*; do btrfs sub delete $f &> /dev/null || true; done
 621             fi
 622             if [ -n "$filestore_path" ]; then
 623                 ln -s $filestore_path $CEPH_DEV_DIR/osd$osd
 624             else
 625                 mkdir -p $CEPH_DEV_DIR/osd$osd
 626             fi
 627
 628             local uuid=`uuidgen`
 629             echo "add osd$osd $uuid"
 630             ceph_adm osd create $uuid
 631             ceph_adm osd crush add osd.$osd 1.0 host=$HOSTNAME root=default
 632             OSD_SECRET=$($CEPH_BIN/ceph-authtool --gen-print-key)
 633             $SUDO $CEPH_BIN/ceph-osd -i $osd $ARGS --mkfs --key $OSD_SECRET --osd-uuid $uuid
 634
 635             local key_fn=$CEPH_DEV_DIR/osd$osd/keyring
 636             cat > $key_fn<<EOF
 637 [osd.$osd]
 638         key = $OSD_SECRET
 639 EOF
 640             echo adding osd$osd key to auth repository
 641             ceph_adm -i "$key_fn" auth add osd.$osd osd "allow *" mon "allow profile osd" mgr "allow profile osd"
 642         fi
 643         echo start osd.$osd
 644         run 'osd' $SUDO $CEPH_BIN/ceph-osd -i $osd $ARGS $COSD_ARGS
 645     done
 646 }
 647
 648 start_mgr() {
 649     local mgr=0
 650     # avoid monitors on nearby ports (which test/*.sh use extensively)
 651     MGR_PORT=$(($CEPH_PORT + 1000))
 652     for name in x y z a b c d e f g h i j k l m n o p
 653     do
 654         [ $mgr -eq $CEPH_NUM_MGR ] && break
 655         mgr=$(($mgr + 1))
 656         if [ "$new" -eq 1 ]; then
 657             mkdir -p $CEPH_DEV_DIR/mgr.$name
 658             key_fn=$CEPH_DEV_DIR/mgr.$name/keyring
 659             $SUDO $CEPH_BIN/ceph-authtool --create-keyring --gen-key --name=mgr.$name $key_fn
 660             ceph_adm -i $key_fn auth add mgr.$name mon 'allow profile mgr' mds 'allow *' osd 'allow *'
 661         fi
 662
 663         wconf <<EOF
 664 [mgr.$name]
 665         host = $HOSTNAME
 666 EOF
 667
 668         ceph_adm config-key set mgr/dashboard/$name/server_port $MGR_PORT
 669         DASH_URLS+="http://$IP:$MGR_PORT/"
 670         MGR_PORT=$(($MGR_PORT + 1000))
 671
 672         ceph_adm config-key set mgr/restful/$name/server_port $MGR_PORT
 673
 674         RESTFUL_URLS+="https://$IP:$MGR_PORT"
 675         MGR_PORT=$(($MGR_PORT + 1000))
 676
 677         echo "Starting mgr.${name}"
 678         run 'mgr' $CEPH_BIN/ceph-mgr -i $name $ARGS
 679     done
 680
 681     # use tell mgr here because the first mgr might not have activated yet
 682     # to register the python module commands.
 683     if ceph_adm tell mgr restful create-self-signed-cert; then
 684         SF=`mktemp`
 685         ceph_adm restful create-key admin -o $SF
 686         RESTFUL_SECRET=`cat $SF`
 687         rm $SF
 688     else
 689         echo MGR Restful is not working, perhaps the package is not installed?
 690     fi
 691 }
 692
 693 start_mds() {
 694     if [ $new -eq 1 ]; then
 695         if [ "$CEPH_NUM_FS" -gt "0" ] ; then
 696             if [ "$CEPH_NUM_FS" -gt "1" ] ; then
 697                 ceph_adm fs flag set enable_multiple true --yes-i-really-mean-it
 698             fi
 699
 700             local fs=0
 701             for name in a b c d e f g h i j k l m n o p
 702             do
 703                 ceph_adm osd pool create "cephfs_data_${name}" 8
 704                 ceph_adm osd pool create "cephfs_metadata_${name}" 8
 705                 ceph_adm fs new "cephfs_${name}" "cephfs_metadata_${name}" "cephfs_data_${name}"
 706                 fs=$(($fs + 1))
 707                 [ $fs -eq $CEPH_NUM_FS ] && break
 708             done
 709         fi
 710     fi
 711
 712     local mds=0
 713     for name in a b c d e f g h i j k l m n o p
 714     do
 715             [ $mds -eq $CEPH_NUM_MDS ] && break
 716             mds=$(($mds + 1))
 717
 718             if [ "$new" -eq 1 ]; then
 719                 prun mkdir -p "$CEPH_DEV_DIR/mds.$name"
 720                 key_fn=$CEPH_DEV_DIR/mds.$name/keyring
 721                 wconf <<EOF
 722 [mds.$name]
 723         host = $HOSTNAME
 724 EOF
 725                     if [ "$standby" -eq 1 ]; then
 726                         mkdir -p $CEPH_DEV_DIR/mds.${name}s
 727                             wconf <<EOF
 728        mds standby for rank = $mds
 729 [mds.${name}s]
 730         mds standby replay = true
 731         mds standby for name = ${name}
 732 EOF
 733                 fi
 734                 prun $SUDO "$CEPH_BIN/ceph-authtool" --create-keyring --gen-key --name="mds.$name" "$key_fn"
 735                 ceph_adm -i "$key_fn" auth add "mds.$name" mon 'allow profile mds' osd 'allow *' mds 'allow' mgr 'allow profile mds'
 736                 if [ "$standby" -eq 1 ]; then
 737                             prun $SUDO "$CEPH_BIN/ceph-authtool" --create-keyring --gen-key --name="mds.${name}s" \
 738                                      "$CEPH_DEV_DIR/mds.${name}s/keyring"
 739                             ceph_adm -i "$CEPH_DEV_DIR/mds.${name}s/keyring" auth add "mds.${name}s" \
 740                                          mon 'allow profile mds' osd 'allow *' mds 'allow' mgr 'allow profile mds'
 741                 fi
 742             fi
 743
 744             run 'mds' $CEPH_BIN/ceph-mds -i $name $ARGS $CMDS_ARGS
 745             if [ "$standby" -eq 1 ]; then
 746                 run 'mds' $CEPH_BIN/ceph-mds -i ${name}s $ARGS $CMDS_ARGS
 747             fi
 748
 749         #valgrind --tool=massif $CEPH_BIN/ceph-mds $ARGS --mds_log_max_segments 2 --mds_thrash_fragments 0 --mds_thrash_exports 0 > m  #--debug_ms 20
 750         #$CEPH_BIN/ceph-mds -d $ARGS --mds_thrash_fragments 0 --mds_thrash_exports 0 #--debug_ms 20
 751         #ceph_adm mds set max_mds 2
 752     done
 753 }
 754
 755 if [ "$debug" -eq 0 ]; then
 756     CMONDEBUG='
 757         debug mon = 10
 758         debug ms = 1'
 759     COSDDEBUG='
 760         debug ms = 1'
 761     CMDSDEBUG='
 762         debug ms = 1'
 763     CMGRDEBUG='
 764         debug ms = 1'
 765 else
 766     echo "** going verbose **"
 767     CMONDEBUG='
 768         debug mon = 20
 769         debug paxos = 20
 770         debug auth = 20
 771         debug mgrc = 20
 772         debug ms = 1'
 773     COSDDEBUG='
 774         debug ms = 1
 775         debug osd = 25
 776         debug objecter = 20
 777         debug monc = 20
 778         debug mgrc = 20
 779         debug journal = 20
 780         debug filestore = 20
 781         debug bluestore = 30
 782         debug bluefs = 20
 783         debug rocksdb = 10
 784         debug bdev = 20
 785         debug rgw = 20
 786         debug reserver = 10
 787         debug objclass = 20'
 788     CMDSDEBUG='
 789         debug ms = 1
 790         debug mds = 20
 791         debug auth = 20
 792         debug monc = 20
 793         debug mgrc = 20
 794         mds debug scatterstat = true
 795         mds verify scatter = true
 796         mds log max segments = 2'
 797     CMGRDEBUG='
 798         debug ms = 1
 799         debug monc = 20
 800         debug mon = 20
 801         debug mgr = 20'
 802 fi
 803
 804 if [ -n "$MON_ADDR" ]; then
 805         CMON_ARGS=" -m "$MON_ADDR
 806         COSD_ARGS=" -m "$MON_ADDR
 807         CMDS_ARGS=" -m "$MON_ADDR
 808 fi
 809
 810 if [ "$memstore" -eq 1 ]; then
 811     COSDMEMSTORE='
 812         osd objectstore = memstore'
 813 fi
 814 if [ "$bluestore" -eq 1 ]; then
 815     COSDMEMSTORE='
 816         osd objectstore = bluestore'
 817 fi
 818
 819 if [ -z "$CEPH_PORT" ]; then
 820     CEPH_PORT=6789
 821     [ -e ".ceph_port" ] && CEPH_PORT=`cat .ceph_port`
 822 fi
 823
 824 [ -z "$INIT_CEPH" ] && INIT_CEPH=$CEPH_BIN/init-ceph
 825
 826 # sudo if btrfs
 827 test -d $CEPH_DEV_DIR/osd0/. && test -e $CEPH_DEV_DIR/sudo && SUDO="sudo"
 828
 829 prun $SUDO rm -f core*
 830
 831 test -d $CEPH_ASOK_DIR || mkdir $CEPH_ASOK_DIR
 832 test -d $CEPH_OUT_DIR || mkdir $CEPH_OUT_DIR
 833 test -d $CEPH_DEV_DIR || mkdir $CEPH_DEV_DIR
 834 $SUDO rm -rf $CEPH_OUT_DIR/*
 835 test -d gmon && $SUDO rm -rf gmon/*
 836
 837 [ "$cephx" -eq 1 ] && [ "$new" -eq 1 ] && test -e $keyring_fn && rm $keyring_fn
 838
 839
 840 # figure machine's ip
 841 HOSTNAME=`hostname -s`
 842 if [ -n "$ip" ]; then
 843     IP="$ip"
 844 else
 845     echo hostname $HOSTNAME
 846     if [ -x "$(which ip 2>/dev/null)" ]; then
 847         IP_CMD="ip addr"
 848     else
 849         IP_CMD="ifconfig"
 850     fi
 851     # filter out IPv6 and localhost addresses
 852     IP="$($IP_CMD | sed -En 's/127.0.0.1//;s/.*inet (addr:)?(([0-9]*\.){3}[0-9]*).*/\2/p' | head -n1)"
 853     # if nothing left, try using localhost address, it might work
 854     if [ -z "$IP" ]; then IP="127.0.0.1"; fi
 855 fi
 856 echo "ip $IP"
 857 echo "port $CEPH_PORT"
 858
 859
 860 [ -z $CEPH_ADM ] && CEPH_ADM=$CEPH_BIN/ceph
 861
 862 ceph_adm() {
 863     if [ "$cephx" -eq 1 ]; then
 864         prun $SUDO "$CEPH_ADM" -c "$conf_fn" -k "$keyring_fn" "$@"
 865     else
 866         prun $SUDO "$CEPH_ADM" -c "$conf_fn" "$@"
 867     fi
 868 }
 869
 870 if [ "$new" -eq 1 ]; then
 871     prepare_conf
 872 fi
 873
 874 if [ $CEPH_NUM_MON -gt 0 ]; then
 875     start_mon
 876 fi
 877
 878 if [ $CEPH_NUM_MGR -gt 0 ]; then
 879     start_mgr
 880 fi
 881
 882 # osd
 883 if [ $CEPH_NUM_OSD -gt 0 ]; then
 884     start_osd
 885 fi
 886
 887 # mds
 888 if [ "$smallmds" -eq 1 ]; then
 889     wconf <<EOF
 890 [mds]
 891         mds log max segments = 2
 892         mds cache size = 10000
 893 EOF
 894 fi
 895
 896 if [ $CEPH_NUM_MDS -gt 0 ]; then
 897     start_mds
 898 fi
 899
 900 # Don't set max_mds until all the daemons are started, otherwise
 901 # the intended standbys might end up in active roles.
 902 if [ "$CEPH_MAX_MDS" -gt 1 ]; then
 903     sleep 5  # wait for daemons to make it into FSMap before increasing max_mds
 904 fi
 905 fs=0
 906 for name in a b c d e f g h i j k l m n o p
 907 do
 908     [ $fs -eq $CEPH_NUM_FS ] && break
 909     fs=$(($fs + 1))
 910     if [ "$CEPH_MAX_MDS" -gt 1 ]; then
 911         ceph_adm fs set "cephfs_${name}" allow_multimds true --yes-i-really-mean-it
 912         ceph_adm fs set "cephfs_${name}" max_mds "$CEPH_MAX_MDS"
 913     fi
 914 done
 915
 916 # mgr
 917
 918 if [ "$ec" -eq 1 ]; then
 919     ceph_adm <<EOF
 920 osd erasure-code-profile set ec-profile m=2 k=2
 921 osd pool create ec 8 8 erasure ec-profile
 922 EOF
 923 fi
 924
 925 do_cache() {
 926     while [ -n "$*" ]; do
 927         p="$1"
 928         shift
 929         echo "creating cache for pool $p ..."
 930         ceph_adm <<EOF
 931 osd pool create ${p}-cache 8
 932 osd tier add $p ${p}-cache
 933 osd tier cache-mode ${p}-cache writeback
 934 osd tier set-overlay $p ${p}-cache
 935 EOF
 936     done
 937 }
 938 do_cache $cache
 939
 940 do_hitsets() {
 941     while [ -n "$*" ]; do
 942         pool="$1"
 943         type="$2"
 944         shift
 945         shift
 946         echo "setting hit_set on pool $pool type $type ..."
 947         ceph_adm <<EOF
 948 osd pool set $pool hit_set_type $type
 949 osd pool set $pool hit_set_count 8
 950 osd pool set $pool hit_set_period 30
 951 EOF
 952     done
 953 }
 954 do_hitsets $hitset
 955
 956 do_rgw_create_users()
 957 {
 958     # Create S3 user
 959     local akey='0555b35654ad1656d804'
 960     local skey='h7GhxuBLTrlhVUyxSPUKUV8r/2EI4ngqJxD7iBdBYLhwluN30JaT3Q=='
 961     echo "setting up user testid"
 962     $CEPH_BIN/radosgw-admin user create --uid testid --access-key $akey --secret $skey --display-name 'M. Tester' --email tester@ceph.com -c $conf_fn > /dev/null
 963
 964     # Create S3-test users
 965     # See: https://github.com/ceph/s3-tests
 966     echo "setting up s3-test users"
 967     $CEPH_BIN/radosgw-admin user create \
 968         --uid 0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef \
 969         --access-key ABCDEFGHIJKLMNOPQRST \
 970         --secret abcdefghijklmnopqrstuvwxyzabcdefghijklmn \
 971         --display-name youruseridhere \
 972         --email s3@example.com -c $conf_fn > /dev/null
 973     $CEPH_BIN/radosgw-admin user create \
 974         --uid 56789abcdef0123456789abcdef0123456789abcdef0123456789abcdef01234 \
 975         --access-key NOPQRSTUVWXYZABCDEFG \
 976         --secret nopqrstuvwxyzabcdefghijklmnabcdefghijklm \
 977         --display-name john.doe \
 978         --email john.doe@example.com -c $conf_fn > /dev/null
 979     $CEPH_BIN/radosgw-admin user create \
 980         --tenant testx \
 981         --uid 9876543210abcdef0123456789abcdef0123456789abcdef0123456789abcdef \
 982         --access-key HIJKLMNOPQRSTUVWXYZA \
 983         --secret opqrstuvwxyzabcdefghijklmnopqrstuvwxyzab \
 984         --display-name tenanteduser \
 985         --email tenanteduser@example.com -c $conf_fn > /dev/null
 986
 987     # Create Swift user
 988     echo "setting up user tester"
 989     $CEPH_BIN/radosgw-admin user create -c $conf_fn --subuser=test:tester --display-name=Tester-Subuser --key-type=swift --secret=testing --access=full > /dev/null
 990
 991     echo ""
 992     echo "S3 User Info:"
 993     echo "  access key:  $akey"
 994     echo "  secret key:  $skey"
 995     echo ""
 996     echo "Swift User Info:"
 997     echo "  account   : test"
 998     echo "  user      : tester"
 999     echo "  password  : testing"
1000     echo ""
1001 }
1002
1003 do_rgw()
1004 {
1005     if [ "$new" -eq 1 ]; then
1006         do_rgw_create_users
1007         if [ -n "$rgw_compression" ]; then
1008             echo "setting compression type=$rgw_compression"
1009             $CEPH_BIN/radosgw-admin zone placement modify -c $conf_fn --rgw-zone=default --placement-id=default-placement --compression=$rgw_compression > /dev/null
1010         fi
1011     fi
1012     # Start server
1013     RGWDEBUG=""
1014     if [ "$debug" -ne 0 ]; then
1015         RGWDEBUG="--debug-rgw=20"
1016     fi
1017
1018     RGWSUDO=
1019     [ $CEPH_RGW_PORT -lt 1024 ] && RGWSUDO=sudo
1020     n=$(($CEPH_NUM_RGW - 1))
1021     i=0
1022     for rgw in j k l m n o p q r s t u v; do
1023         echo start rgw on http://localhost:$((CEPH_RGW_PORT + i))
1024         run 'rgw' $RGWSUDO $CEPH_BIN/radosgw -c $conf_fn --log-file=${CEPH_OUT_DIR}/rgw.$rgw.log ${RGWDEBUG} --debug-ms=1 -n client.rgw "--rgw_frontends=${rgw_frontend} port=$((CEPH_RGW_PORT + i))"
1025         i=$(($i + 1))
1026         [ $i -eq $CEPH_NUM_RGW ] && break
1027     done
1028 }
1029 if [ "$CEPH_NUM_RGW" -gt 0 ]; then
1030     do_rgw
1031 fi
1032
1033 echo "started.  stop.sh to stop.  see out/* (e.g. 'tail -f out/????') for debug output."
1034
1035 echo ""
1036 echo "dashboard urls: $DASH_URLS"
1037 echo "  restful urls: $RESTFUL_URLS"
1038 echo "  w/ user/pass: admin / $RESTFUL_SECRET"
1039 echo ""
1040 echo "export PYTHONPATH=./pybind:$PYTHONPATH"
1041 echo "export LD_LIBRARY_PATH=$CEPH_LIB"
1042
1043 if [ "$CEPH_DIR" != "$PWD" ]; then
1044     echo "export CEPH_CONF=$conf_fn"
1045     echo "export CEPH_KEYRING=$keyring_fn"
1046 fi
1047
1048 echo "CEPH_DEV=1"
1049