gp.py 51.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
#!/usr/bin/env python
#
# Copyright (c) Greenplum Inc 2008. All Rights Reserved.
#

"""
TODO: docs!
"""
import os, pickle, base64, time

J
Jamie McAtamney 已提交
11
import re, socket
L
Larry Hamel 已提交
12

13 14 15
from gppylib.gplog import *
from gppylib.db import dbconn
from gppylib import gparray
L
Larry Hamel 已提交
16
from gppylib.commands.base import *
17 18 19
from unix import *
import pg
from gppylib import pgconf
20
from gppylib.utils import writeLinesToFile, createFromSingleHostFile, shellEscape
21 22 23 24 25 26 27 28 29 30 31 32 33 34


logger = get_default_logger()

#TODO:  need a better way of managing environment variables.
GPHOME=os.environ.get('GPHOME')

#Default timeout for segment start
SEGMENT_TIMEOUT_DEFAULT=600
SEGMENT_STOP_TIMEOUT_DEFAULT=120

#"Command not found" return code in bash
COMMAND_NOT_FOUND=127

35 36 37
#Default size of thread pool for gpstart and gpsegstart
DEFAULT_GPSTART_NUM_WORKERS=64

38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
def get_postmaster_pid_locally(datadir):
    cmdStr = "ps -ef | grep postgres | grep -v grep | awk '{print $2}' | grep `cat %s/postmaster.pid | head -1` || echo -1" % (datadir)
    name = "get postmaster"
    cmd = Command(name, cmdStr)
    try:
        cmd.run(validateAfter=True)
        sout = cmd.get_results().stdout.lstrip(' ')
        return int(sout.split()[0])
    except:
        return -1

def getPostmasterPID(hostname, datadir):
    cmdStr="ps -ef | grep postgres | grep -v grep | awk '{print $2}' | grep \\`cat %s/postmaster.pid | head -1\\` || echo -1" % (datadir)
    name="get postmaster pid"
    cmd=Command(name,cmdStr,ctxt=REMOTE,remoteHost=hostname)
    try:
        cmd.run(validateAfter=True)
        sout=cmd.get_results().stdout.lstrip(' ')
        return int(sout.split()[1])
    except:
        return -1

#-----------------------------------------------

class CmdArgs(list):
    """
M
Marbin Tan 已提交
64 65
    Conceptually this is a list of an executable path and executable options
    built in a structured manner with a canonical string representation suitable
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
    for execution via a shell.

    Examples
    --------

    >>> str(CmdArgs(['foo']).set_verbose(True))
    'foo -v'
    >>> str(CmdArgs(['foo']).set_verbose(False).set_wait_timeout(True,600))
    'foo -w -t 600'

    """

    def __init__(self, l):
        list.__init__(self, l)

    def __str__(self):
        return " ".join(self)

    def set_verbose(self, verbose):
        """
        @param verbose - true if verbose output desired
        """
        if verbose: self.append("-v")
        return self

    def set_wrapper(self, wrapper, args):
        """
        @param wrapper - wrapper executable ultimately passed to pg_ctl
        @param args - wrapper arguments ultimately passed to pg_ctl
        """
        if wrapper:
            self.append("--wrapper=\"%s\"" % wrapper)
            if args:
                self.append("--wrapper-args=\"%s\"" % args)
        return self

    def set_wait_timeout(self, wait, timeout):
        """
        @param wait: true if should wait until operation completes
        @param timeout: number of seconds to wait before giving up
        """
M
Marbin Tan 已提交
107
        if wait:
108
            self.append("-w")
M
Marbin Tan 已提交
109
        if timeout:
110 111 112 113 114 115 116 117 118
            self.append("-t")
            self.append(str(timeout))
        return self

    def set_segments(self, segments):
        """
        @param segments - segments (from GpArray.getSegmentsByHostName)
        """
        for seg in segments:
H
Heikki Linnakangas 已提交
119 120
            cfg_array = repr(seg)
            self.append("-D '%s'" % (cfg_array))
121 122 123 124 125 126 127
        return self



class PgCtlBackendOptions(CmdArgs):
    """
    List of options suitable for use with the -o option of pg_ctl.
M
Marbin Tan 已提交
128
    Used by MasterStart, SegmentStart to format the backend options
129 130 131 132 133 134
    string passed via pg_ctl -o

    Examples
    --------

    >>> str(PgCtlBackendOptions(5432, 1, 2))
135
    '-p 5432 --silent-mode=true'
136
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_master(True))
137
    '-p 5432 --silent-mode=true -i'
D
David Kimura 已提交
138
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_master(False))
139
    '-p 5432 --silent-mode=true -i -E'
140
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_segment(1))
141
    '-p 5432 --silent-mode=true -i'
142
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_special('upgrade'))
143
    '-p 5432 --silent-mode=true -U'
144
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_special('maintenance'))
145
    '-p 5432 --silent-mode=true -m'
146
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_utility(True))
147
    '-p 5432 --silent-mode=true -c gp_role=utility'
148
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_utility(False))
149
    '-p 5432 --silent-mode=true'
150
    >>> str(PgCtlBackendOptions(5432, 1, 2).set_restricted(True,1))
151
    '-p 5432 --silent-mode=true -c superuser_reserved_connections=1'
M
Marbin Tan 已提交
152
    >>>
153 154 155

    """

156
    def __init__(self, port):
157 158 159 160 161 162 163 164 165 166 167
        """
        @param port: backend port
        """
        CmdArgs.__init__(self, [
            "-p", str(port),
        ])

    #
    # master/segment-specific options
    #

D
David Kimura 已提交
168
    def set_master(self, is_utility_mode):
169
        """
D
David Kimura 已提交
170
        @param is_utility_mode: start with is_utility_mode?
171
        """
D
David Kimura 已提交
172
        if not is_utility_mode: self.append("-E")
173 174 175 176 177 178 179 180
        return self

    #
    # startup mode options
    #

    def set_special(self, special):
        """
181
        @param special: special mode (none, 'upgrade', 'maintenance', 'convertMasterDataDirToSegment')
182
        """
183
        opt = {None:None, 'upgrade':'-U', 'maintenance':'-m', 'convertMasterDataDirToSegment':'-M'}[special]
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
        if opt: self.append(opt)
        return self

    def set_utility(self, utility):
        """
        @param utility: true if starting in utility mode
        """
        if utility: self.append("-c gp_role=utility")
        return self

    def set_restricted(self, restricted, max_connections):
        """
        @param restricted: true if restricting connections
        @param max_connections: connection limit
        """
        if restricted:  self.append("-c superuser_reserved_connections=%s" % max_connections)
        return self


class PgCtlStartArgs(CmdArgs):
    """
    Used by MasterStart, SegmentStart to format the pg_ctl command
    to start a backend postmaster.

    Examples
    --------

    >>> a = PgCtlStartArgs("/data1/master/gpseg-1", str(PgCtlBackendOptions(5432, 1, 2)), 123, None, None, True, 600)
    >>> str(a).split(' ') #doctest: +NORMALIZE_WHITESPACE
M
Marbin Tan 已提交
213 214
    ['env', GPERA=123', '$GPHOME/bin/pg_ctl', '-D', '/data1/master/gpseg-1', '-l',
     '/data1/master/gpseg-1/pg_log/startup.log', '-w', '-t', '600',
215
     '-o', '"', '-p', '5432', '--silent-mode=true', '"', 'start']
216 217 218 219 220 221 222 223 224 225 226 227 228 229
    """

    def __init__(self, datadir, backend, era, wrapper, args, wait, timeout=None):
        """
        @param datadir: database data directory
        @param backend: backend options string from PgCtlBackendOptions
        @param era: gpdb master execution era
        @param wrapper: wrapper executable for pg_ctl
        @param args: wrapper arguments for pg_ctl
        @param wait: true if pg_ctl should wait until backend starts completely
        @param timeout: number of seconds to wait before giving up
        """

        CmdArgs.__init__(self, [
L
Larry Hamel 已提交
230
            "env",
231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
            "GPSESSID=0000000000", 	# <- overwritten with gp_session_id to help identify orphans
            "GPERA=%s" % str(era),	# <- master era used to help identify orphans
            "$GPHOME/bin/pg_ctl",
            "-D", str(datadir),
            "-l", "%s/pg_log/startup.log" % datadir,
        ])
        self.set_wrapper(wrapper, args)
        self.set_wait_timeout(wait, timeout)
        self.extend([
            "-o", "\"", str(backend), "\"",
            "start"
        ])


class PgCtlStopArgs(CmdArgs):
    """
    Used by MasterStop, SegmentStop to format the pg_ctl command
    to stop a backend postmaster

    >>> str(PgCtlStopArgs("/data1/master/gpseg-1", "smart", True, 600))
    '$GPHOME/bin/pg_ctl -D /data1/master/gpseg-1 -m smart -w -t 600 stop'

    """

    def __init__(self, datadir, mode, wait, timeout):
        """
        @param datadir: database data directory
        @param mode: shutdown mode (smart, fast, immediate)
        @param wait: true if pg_ctlshould wait for backend to stop
        @param timeout: number of seconds to wait before giving up
        """
        CmdArgs.__init__(self, [
            "$GPHOME/bin/pg_ctl",
            "-D", str(datadir),
            "-m", str(mode),
        ])
        self.set_wait_timeout(wait, timeout)
        self.append("stop")


class MasterStart(Command):
272
    def __init__(self, name, dataDir, port, era,
273
                 wrapper, wrapper_args, specialMode=None, restrictedMode=False, timeout=SEGMENT_TIMEOUT_DEFAULT,
274
                 max_connections=1, utilityMode=False, ctxt=LOCAL, remoteHost=None,
275 276 277 278 279 280 281 282 283
                 wait=True
                 ):
        self.dataDir=dataDir
        self.port=port
        self.utilityMode=utilityMode
        self.wrapper=wrapper
        self.wrapper_args=wrapper_args

        # build backend options
284
        b = PgCtlBackendOptions(port)
D
David Kimura 已提交
285
        b.set_master(is_utility_mode=utilityMode)
286 287 288 289 290 291 292 293 294 295 296
        b.set_utility(utilityMode)
        b.set_special(specialMode)
        b.set_restricted(restrictedMode, max_connections)

        # build pg_ctl command
        c = PgCtlStartArgs(dataDir, b, era, wrapper, wrapper_args, wait, timeout)
        self.cmdStr = str(c)

        Command.__init__(self, name, self.cmdStr, ctxt, remoteHost)

    @staticmethod
297
    def local(name, dataDir, port, era,
298
              wrapper, wrapper_args, specialMode=None, restrictedMode=False, timeout=SEGMENT_TIMEOUT_DEFAULT,
299
              max_connections=1, utilityMode=False):
300
        cmd=MasterStart(name, dataDir, port, era,
301
                        wrapper, wrapper_args, specialMode, restrictedMode, timeout,
302
                        max_connections, utilityMode)
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325
        cmd.run(validateAfter=True)

#-----------------------------------------------
class MasterStop(Command):
    def __init__(self,name,dataDir,mode='smart',timeout=SEGMENT_STOP_TIMEOUT_DEFAULT, ctxt=LOCAL,remoteHost=None):
        self.dataDir = dataDir
        self.cmdStr = str( PgCtlStopArgs(dataDir, mode, True, timeout) )
        Command.__init__(self, name, self.cmdStr, ctxt, remoteHost)

    @staticmethod
    def local(name,dataDir):
        cmd=MasterStop(name,dataDir)
        cmd.run(validateAfter=True)

#-----------------------------------------------
class SegmentStart(Command):
    """
    SegmentStart is used to start a single segment.

    Note: Most code should probably use GpSegStartCmd instead which starts up
    all of the segments on a specified GpHost.
    """

M
Mike Roth 已提交
326
    def __init__(self, name, gpdb, numContentsInCluster, era, mirrormode,
327
                 utilityMode=False, ctxt=LOCAL, remoteHost=None,
S
Shoaib Lari 已提交
328
                 pg_ctl_wait=True, timeout=SEGMENT_TIMEOUT_DEFAULT,
329 330 331 332 333 334 335 336 337 338
                 specialMode=None, wrapper=None, wrapper_args=None):

        # This is referenced from calling code
        self.segment = gpdb

        # Interesting data from our input segment
        port    = gpdb.getSegmentPort()
        datadir = gpdb.getSegmentDataDirectory()

        # build backend options
339
        b = PgCtlBackendOptions(port)
340 341 342 343
        b.set_utility(utilityMode)
        b.set_special(specialMode)

        # build pg_ctl command
S
Shoaib Lari 已提交
344
        c = PgCtlStartArgs(datadir, b, era, wrapper, wrapper_args, pg_ctl_wait, timeout)
345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
        self.cmdStr = str(c) + ' 2>&1'

        Command.__init__(self, name, self.cmdStr, ctxt, remoteHost)

    @staticmethod
    def local(name, gpdb, numContentsInCluster, era, mirrormode, utilityMode=False):
        cmd=SegmentStart(name, gpdb, numContentsInCluster, era, mirrormode, utilityMode)
        cmd.run(validateAfter=True)

    @staticmethod
    def remote(name, remoteHost, gpdb, numContentsInCluster, era, mirrormode, utilityMode=False):
        cmd=SegmentStart(name, gpdb, numContentsInCluster, era, mirrormode, utilityMode, ctxt=REMOTE, remoteHost=remoteHost)
        cmd.run(validateAfter=True)

#-----------------------------------------------
class SegmentStop(Command):
M
Marbin Tan 已提交
361
    def __init__(self, name, dataDir,mode='smart', nowait=False, ctxt=LOCAL,
362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398
                 remoteHost=None, timeout=SEGMENT_STOP_TIMEOUT_DEFAULT):

        self.cmdStr = str( PgCtlStopArgs(dataDir, mode, not nowait, timeout) )
        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)

    @staticmethod
    def local(name, dataDir,mode='smart'):
        cmd=SegmentStop(name, dataDir,mode)
        cmd.run(validateAfter=True)
        return cmd

    @staticmethod
    def remote(name, hostname, dataDir, mode='smart'):
        cmd=SegmentStop(name, dataDir, mode, ctxt=REMOTE, remoteHost=hostname)
        cmd.run(validateAfter=True)
        return cmd

#-----------------------------------------------
class SegmentIsShutDown(Command):
    """
    Get the pg_controldata status, and check that it says 'shut down'
    """
    def __init__(self,name,directory,ctxt=LOCAL,remoteHost=None):
        cmdStr = "$GPHOME/bin/pg_controldata %s" % directory
        Command.__init__(self,name,cmdStr,ctxt,remoteHost)

    def is_shutdown(self):
        for key, value in self.results.split_stdout():
            if key == 'Database cluster state':
                return value.strip() == 'shut down'
        return False

    @staticmethod
    def local(name,directory):
        cmd=SegmentIsShutDown(name,directory)
        cmd.run(validateAfter=True)

399 400 401 402 403 404 405
#-----------------------------------------------
class SegmentRewind(Command):
    """
    SegmentRewind is used to run pg_rewind using source server.
    """

    def __init__(self, name, target_host, target_datadir,
406 407
                 source_host, source_port,
                 verbose=False, ctxt=REMOTE):
408 409 410 411 412 413 414

        # Construct the source server libpq connection string
        source_server = "host=%s port=%s dbname=template1" % (source_host, source_port)

        # Build the pg_rewind command. Do not run pg_rewind if recovery.conf
        # file exists in target data directory because the target instance can
        # be started up normally as a mirror for WAL replication catch up.
A
Alexandra Wang 已提交
415
        rewind_cmd = '[ -f %s/recovery.conf ] || PGOPTIONS="-c gp_session_role=utility" $GPHOME/bin/pg_rewind --write-recovery-conf --slot="internal_wal_replication_slot" --source-server="%s" --target-pgdata=%s' % (target_datadir, source_server, target_datadir)
416

417 418 419
        if verbose:
            rewind_cmd = rewind_cmd + ' --progress'

420 421 422
        self.cmdStr = rewind_cmd + ' 2>&1'

        Command.__init__(self, name, self.cmdStr, ctxt, target_host)
423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511

#
# list of valid segment statuses that can be requested
#

SEGMENT_STATUS_GET_STATUS = "getStatus"

#
# corresponds to a postmaster string value; result is a string object, or None if version could not be fetched
#
SEGMENT_STATUS__GET_VERSION = "getVersion"

#
# corresponds to a postmaster string value; result is a dictionary object, or None if data could not be fetched
#
# dictionary will contain:
#    mode -> string
#    segmentState -> string
#    dataState -> string
#    resyncNumCompleted -> large integer
#    resyncTotalToComplete -> large integer
#    elapsedTimeSeconds -> large integer
#
SEGMENT_STATUS__GET_MIRROR_STATUS = "getMirrorStatus"

#
# fetch the active PID of this segment; result is a dict with "pid" and "error" values
#
# see comments on getPidStatus in GpSegStatusProgram class
#
SEGMENT_STATUS__GET_PID = "__getPid"

#
# fetch True or False depending on whether the /tmp/.s.PSQL.<port>.lock file is there
#
SEGMENT_STATUS__HAS_LOCKFILE = "__hasLockFile"

#
# fetch True or False depending on whether the postmaster pid file is there
#
SEGMENT_STATUS__HAS_POSTMASTER_PID_FILE = "__hasPostmasterPidFile"

class GpGetStatusUsingTransitionArgs(CmdArgs):
    """
    Examples
    --------

    >>> str(GpGetStatusUsingTransitionArgs([],'request'))
    '$GPHOME/sbin/gpgetstatususingtransition.py -s request'
    """

    def __init__(self, segments, status_request):
        """
        @param status_request
        """
        CmdArgs.__init__(self, [
            "$GPHOME/sbin/gpgetstatususingtransition.py",
            "-s", str(status_request)
        ])
        self.set_segments(segments)


class GpGetSegmentStatusValues(Command):
    """
    Fetch status values for segments on a host

    Results will be a bin-hexed/pickled value that, when unpacked, will give a
    two-level map:

    outer-map maps from SEGMENT_STATUS__* value to inner-map
    inner-map maps from dbid to result (which is usually a string, but may be different)

    @param statusRequestArr an array of SEGMENT_STATUS__ constants
    """
    def __init__(self, name, segments, statusRequestArr, verbose=False, ctxt=LOCAL, remoteHost=None):

        # clone the list
        self.dblist = [x for x in segments]

        # build gpgetstatususingtransition commadn
        status_request = ":".join(statusRequestArr)
        c = GpGetStatusUsingTransitionArgs(segments, status_request)
        c.set_verbose(verbose)
        cmdStr = str(c)

        Command.__init__(self, name, cmdStr, ctxt, remoteHost)

    def decodeResults(self):
        """
M
Marbin Tan 已提交
512
        return (warning,outputFromCmd) tuple, where if warning is None then
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535
           results were returned and outputFromCmd should be read.  Otherwise, the warning should
           be logged and outputFromCmd ignored
        """
        if self.get_results().rc != 0:
            return ("Error getting status from host %s" % self.remoteHost, None)

        outputFromCmd = None
        for line in self.get_results().stdout.split('\n'):
            if line.startswith("STATUS_RESULTS:"):
                toDecode = line[len("STATUS_RESULTS:"):]
                outputFromCmd = pickle.loads(base64.urlsafe_b64decode(toDecode))
                break
        if outputFromCmd is None:
            return ("No status output provided from host %s" % self.remoteHost, None)
        return (None, outputFromCmd)


SEGSTART_ERROR_UNKNOWN_ERROR = -1
SEGSTART_SUCCESS = 0
SEGSTART_ERROR_STOP_RUNNING_SEGMENT_FAILED = 5
SEGSTART_ERROR_DATA_DIRECTORY_DOES_NOT_EXIST = 6
SEGSTART_ERROR_PG_CTL_FAILED = 8
SEGSTART_ERROR_PING_FAILED = 10 # not actually done inside GpSegStartCmd, done instead by caller
S
Shoaib Lari 已提交
536 537
SEGSTART_ERROR_PG_CONTROLDATA_FAILED = 11
SEGSTART_ERROR_CHECKSUM_MISMATCH = 12
M
Marbin Tan 已提交
538

539 540 541 542 543 544 545

class GpSegStartArgs(CmdArgs):
    """
    Examples
    --------

    >>> str(GpSegStartArgs('en_US.utf-8:en_US.utf-8:en_US.utf-8', 'mirrorless', 'gpversion', 1, 123, 600))
546
    "$GPHOME/sbin/gpsegstart.py -M mirrorless -V 'gpversion' -n 1 --era 123 -t 600"
547 548
    """

S
Shoaib Lari 已提交
549
    def __init__(self, mirrormode, gpversion, num_cids, era, master_checksum_value, timeout):
550 551 552 553 554 555 556
        """
        @param mirrormode - mirror start mode (START_AS_PRIMARY_OR_MIRROR or START_AS_MIRRORLESS)
        @param gpversion - version (from postgres --gp-version)
        @param num_cids - number content ids
        @param era - master era
        @param timeout - seconds to wait before giving up
        """
S
Shoaib Lari 已提交
557
        default_args = [
558 559 560 561 562 563
            "$GPHOME/sbin/gpsegstart.py",
            "-M", str(mirrormode),
            "-V '%s'" % gpversion,
            "-n", str(num_cids),
            "--era", str(era),
            "-t", str(timeout)
S
Shoaib Lari 已提交
564 565 566 567 568 569
        ]
        if master_checksum_value != None:
            default_args.append("--master-checksum-version")
            default_args.append(str(master_checksum_value))

        CmdArgs.__init__(self, default_args)
570 571 572 573 574 575

    def set_special(self, special):
        """
        @param special - special mode
        """
        assert(special in [None, 'upgrade', 'maintenance'])
M
Marbin Tan 已提交
576
        if special:
577 578 579 580 581 582 583 584 585 586 587 588 589
            self.append("-U")
            self.append(special)
        return self

    def set_transition(self, data):
        """
        @param data - pickled transition data
        """
        if data is not None:
            self.append("-p")
            self.append(data)
        return self

590 591 592 593 594 595 596 597 598
    def set_parallel(self, parallel):
        """
        @param parallel - maximum size of a thread pool to start segments
        """
        if parallel is not None:
            self.append("-B")
            self.append(str(parallel))
        return self

599 600 601


class GpSegStartCmd(Command):
602
    def __init__(self, name, gphome, segments, gpversion,
S
Shoaib Lari 已提交
603
                 mirrormode, numContentsInCluster, era, master_checksum_value=None,
M
Marbin Tan 已提交
604
                 timeout=SEGMENT_TIMEOUT_DEFAULT, verbose=False,
605
                 ctxt=LOCAL, remoteHost=None, pickledTransitionData=None,
606
                 specialMode=None, wrapper=None, wrapper_args=None,
607
                 parallel=None, logfileDirectory=False):
608 609 610 611 612

        # Referenced by calling code (in operations/startSegments.py), create a clone
        self.dblist = [x for x in segments]

        # build gpsegstart command string
S
Shoaib Lari 已提交
613
        c = GpSegStartArgs(mirrormode, gpversion, numContentsInCluster, era, master_checksum_value,timeout)
614 615 616 617 618
        c.set_verbose(verbose)
        c.set_special(specialMode)
        c.set_transition(pickledTransitionData)
        c.set_wrapper(wrapper, wrapper_args)
        c.set_segments(segments)
619
        c.set_parallel(parallel)
620 621 622 623

        cmdStr = str(c)
        logger.debug(cmdStr)

624 625
        if (logfileDirectory):
            cmdStr = cmdStr + " -l '" + logfileDirectory + "'"
626 627 628 629 630 631
        Command.__init__(self,name,cmdStr,ctxt,remoteHost)


#-----------------------------------------------
class GpSegStopCmd(Command):
    def __init__(self, name, gphome, version,mode,dbs,timeout=SEGMENT_STOP_TIMEOUT_DEFAULT,
632
                 verbose=False, ctxt=LOCAL, remoteHost=None, logfileDirectory=False):
633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
        self.gphome=gphome
        self.dblist=dbs
        self.dirportlist=[]
        self.mode=mode
        self.version=version
        for db in dbs:
            datadir = db.getSegmentDataDirectory()
            port = db.getSegmentPort()
            self.dirportlist.append(datadir + ':' + str(port))
        self.timeout=timeout
        dirstr=" -D ".join(self.dirportlist)
        if verbose:
            setverbose=" -v "
        else:
            setverbose=""

        self.cmdStr="$GPHOME/sbin/gpsegstop.py %s -D %s -m %s -t %s -V '%s'"  %\
                        (setverbose,dirstr,mode,timeout,version)

652 653
        if (logfileDirectory):
            self.cmdStr = self.cmdStr + " -l '" + logfileDirectory + "'"
654 655 656 657 658 659 660 661 662 663 664
        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)


#-----------------------------------------------
class GpStandbyStart(MasterStart, object):
    """
    Start up the master standby.  The options to postgres in standby
    are almost same as primary master, with a few exceptions.
    The standby will be up as dispatch mode, and could be in remote.
    """

665 666
    def __init__(self, name, datadir, port, ctxt=LOCAL,
                 remoteHost=None, era=None,
667 668 669 670 671 672 673 674 675 676 677 678 679 680
                 wrapper=None, wrapper_args=None):
        super(GpStandbyStart, self).__init__(
                name=name,
                dataDir=datadir,
                port=port,
                era=era,
                wrapper=wrapper,
                wrapper_args=wrapper_args,
                ctxt=ctxt,
                remoteHost=remoteHost,
                wait=False
                )

    @staticmethod
681
    def local(name, datadir, port, era=None,
682
              wrapper=None, wrapper_args=None):
683 684
        cmd = GpStandbyStart(name, datadir, port,
                             era=era,
685 686 687 688 689
                             wrapper=wrapper, wrapper_args=wrapper_args)
        cmd.run(validateAfter=True)
        return cmd

    @staticmethod
690
    def remote(name, host, datadir, port, era=None,
691
               wrapper=None, wrapper_args=None):
692 693
        cmd = GpStandbyStart(name, datadir, port, ctxt=REMOTE,
                             remoteHost=host, era=era,
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
                             wrapper=wrapper, wrapper_args=wrapper_args)
        cmd.run(validateAfter=True)
        return cmd

#-----------------------------------------------
class GpStart(Command):
    def __init__(self, name, masterOnly=False, restricted=False, verbose=False,ctxt=LOCAL, remoteHost=None):
        self.cmdStr="$GPHOME/bin/gpstart -a"
        if masterOnly:
            self.cmdStr += " -m"
            self.propagate_env_map['GPSTART_INTERNAL_MASTER_ONLY'] = 1
        if restricted:
            self.cmdStr += " -R"
        if verbose or logging_is_verbose():
            self.cmdStr += " -v"
        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)

    @staticmethod
    def local(name,masterOnly=False,restricted=False):
        cmd=GpStart(name,masterOnly,restricted)
        cmd.run(validateAfter=True)

#-----------------------------------------------
class NewGpStart(Command):
    def __init__(self, name, masterOnly=False, restricted=False, verbose=False,nostandby=False,ctxt=LOCAL, remoteHost=None, masterDirectory=None):
        self.cmdStr="$GPHOME/bin/gpstart -a"
        if masterOnly:
            self.cmdStr += " -m"
            self.propagate_env_map['GPSTART_INTERNAL_MASTER_ONLY'] = 1
        if restricted:
            self.cmdStr += " -R"
        if verbose or logging_is_verbose():
            self.cmdStr += " -v"
        if nostandby:
            self.cmdStr += " -y"
        if masterDirectory:
            self.cmdStr += " -d " + masterDirectory

        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)

    @staticmethod
    def local(name,masterOnly=False,restricted=False,verbose=False,nostandby=False,
              masterDirectory=None):
        cmd=NewGpStart(name,masterOnly,restricted,verbose,nostandby,
                       masterDirectory=masterDirectory)
        cmd.run(validateAfter=True)

#-----------------------------------------------
class NewGpStop(Command):
    def __init__(self, name, masterOnly=False, restart=False, fast=False, force=False, verbose=False, ctxt=LOCAL, remoteHost=None):
        self.cmdStr="$GPHOME/bin/gpstop -a"
        if masterOnly:
            self.cmdStr += " -m"
        if verbose or logging_is_verbose():
            self.cmdStr += " -v"
        if fast:
            self.cmdStr += " -f"
        if restart:
            self.cmdStr += " -r"
        if force:
            self.cmdStr += " -M immediate"
        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)

    @staticmethod
    def local(name,masterOnly=False, restart=False, fast=False, force=False, verbose=False):
        cmd=NewGpStop(name,masterOnly,restart, fast, force, verbose)
        cmd.run(validateAfter=True)

#-----------------------------------------------
class GpStop(Command):
764
    def __init__(self, name, masterOnly=False, verbose=False, quiet=False, restart=False, fast=False, force=False, datadir=None, parallel=None, reload=False, ctxt=LOCAL, remoteHost=None, logfileDirectory=False):
765 766 767 768 769 770 771 772 773 774 775 776 777 778 779
        self.cmdStr="$GPHOME/bin/gpstop -a"
        if masterOnly:
            self.cmdStr += " -m"
        if restart:
            self.cmdStr += " -r"
        if fast:
            self.cmdStr += " -f"
        if force:
            self.cmdStr += " -M immediate"
        if datadir:
            self.cmdStr += " -d %s" % datadir
        if verbose or logging_is_verbose():
            self.cmdStr += " -v"
        if quiet:
            self.cmdStr += " -q"
780 781
        if parallel:
            self.cmdStr += " -B %s" % parallel
782 783
        if logfileDirectory:
            self.cmdStr += " -l '" + logfileDirectory + "'"
784 785
        if reload:
            self.cmdStr += " -u"
786 787 788
        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)

    @staticmethod
789 790
    def local(name,masterOnly=False, verbose=False, quiet=False,restart=False, fast=False, force=False, datadir=None, parallel=None, reload=False):
        cmd=GpStop(name,masterOnly,verbose,quiet,restart,fast,force,datadir,parallel,reload)
791 792 793 794
        cmd.run(validateAfter=True)
        return cmd

#-----------------------------------------------
795
class ModifyConfSetting(Command):
796 797 798 799 800 801 802
    def __init__(self, name, file, optName, optVal, optType='string', ctxt=LOCAL, remoteHost=None):
        cmdStr = None
        if optType == 'number':
            cmdStr = "perl -p -i.bak -e 's/^%s[ ]*=[ ]*\\d+/%s=%d/' %s" % (optName, optName, optVal, file)
        elif optType == 'string':
            cmdStr = "perl -i -p -e \"s/^%s[ ]*=[ ]*'[^']*'/%s='%s'/\" %s" % (optName, optName, optVal, file)
        else:
803
            raise Exception, "Invalid optType for ModifyConfSetting"
804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
        self.cmdStr = cmdStr
        Command.__init__(self, name, self.cmdStr, ctxt, remoteHost)

#-----------------------------------------------
class GpCleanSegmentDirectories(Command):
    """
    Clean all of the directories for a set of segments on the host.

    Does NOT delete all files in the data directories -- tries to preserve logs and any non-database
       files the user has placed there
    """
    def __init__(self, name, segmentsToClean, ctxt, remoteHost):
        pickledSegmentsStr = base64.urlsafe_b64encode(pickle.dumps(segmentsToClean))
        cmdStr = "$GPHOME/sbin/gpcleansegmentdir.py -p %s" % pickledSegmentsStr
        Command.__init__(self, name, cmdStr, ctxt, remoteHost)

#-----------------------------------------------
821
class GpDirsExist(Command):
822 823 824
    """
    Checks if gp_dump* directories exist in the given directory
    """
825 826
    def __init__(self, name, baseDir, dirName, ctxt=LOCAL, remoteHost=None):
        cmdStr = "find %s -name %s -print" % (baseDir, dirName)
827
        Command.__init__(self, name, cmdStr, ctxt, remoteHost)
M
Marbin Tan 已提交
828

829
    @staticmethod
830 831
    def local(name, baseDir, dirName):
        cmd = GpDirsExist(name, baseDir=baseDir, dirName=dirName)
832 833 834 835 836 837 838 839 840 841 842 843 844 845
        cmd.run(validateAfter=True)
        dirCount = len(cmd.get_results().stdout.split('\n'))
        # This is > 1 because the command output will terminate with \n
        return dirCount > 1


#-----------------------------------------------
class ConfigureNewSegment(Command):
    """
    Configure a new segment, usually from a template, as is done during gpexpand, gpaddmirrors, gprecoverseg (full),
      etc.
    """

    def __init__(self, name, confinfo, newSegments=False, tarFile=None,
846 847
                 batchSize=None, verbose=False,ctxt=LOCAL, remoteHost=None, validationOnly=False, writeGpIdFileOnly=False,
                 forceoverwrite=False):
848 849 850 851 852 853 854 855 856 857 858 859 860
        cmdStr = '$GPHOME/bin/lib/gpconfigurenewsegment -c \"%s\"' % (confinfo)
        if newSegments:
            cmdStr += ' -n'
        if tarFile:
            cmdStr += ' -t %s' % tarFile
        if verbose:
            cmdStr += ' -v '
        if batchSize:
            cmdStr += ' -B %s' % batchSize
        if validationOnly:
            cmdStr += " --validation-only"
        if writeGpIdFileOnly:
            cmdStr += " --write-gpid-file-only"
861 862
        if forceoverwrite:
            cmdStr += " --force-overwrite"
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878

        Command.__init__(self, name, cmdStr, ctxt, remoteHost)

    #-----------------------------------------------
    @staticmethod
    def buildSegmentInfoForNewSegment(segments, isTargetReusedLocationArr = None, primaryMirror = 'both'):
        """
        Build the new segment info that can be used to get the confinfo argument to pass to ConfigureNewSegment

        @param segments list of segments

        @param isTargetReusedLocationArr if not None, then is an array of boolean values in parallel with segments
                                      True values indicate that the directory has been cleaned by gpcleansegmentdir.py
                                      and we should have lighter restrictions on how to check it for emptiness
                                      Passing None is the same as passing an array of all False values

M
Marbin Tan 已提交
879 880
        @param primaryMirror Process 'primary' or 'mirror' or 'both'

881
        @return A dictionary with the following format:
M
Marbin Tan 已提交
882

883 884 885 886 887 888 889 890
                Name  =   <host name>
                Value =   <system data directory>
                        : <port>
                        : if primary then 'true' else 'false'
                        : if target is reused location then 'true' else 'false'
                        : <segment dbid>
        """
        result = {}
891

892 893 894 895 896 897 898 899 900 901 902
        for segIndex, seg in enumerate(segments):
            if primaryMirror == 'primary' and seg.isSegmentPrimary() == False:
               continue
            elif primaryMirror == 'mirror' and seg.isSegmentPrimary() == True:
               continue
            hostname = seg.getSegmentHostName()
            if result.has_key(hostname):
                result[hostname] += ','
            else:
                result[hostname] = ''

S
Shoaib Lari 已提交
903
            isTargetReusedLocation = isTargetReusedLocationArr and isTargetReusedLocationArr[segIndex]
904 905 906 907
            # only a mirror segment has these two attributes
            # added on the fly, by callers
            primaryHostname = getattr(seg, 'primaryHostname', "")
            primarySegmentPort = getattr(seg, 'primarySegmentPort', "-1")
S
Shoaib Lari 已提交
908 909 910 911 912 913
            if primaryHostname == "":
                isPrimarySegment =  "true" if seg.isSegmentPrimary(current_role=True) else "false"
                isTargetReusedLocationString = "true" if isTargetReusedLocation else "false"
            else:
                isPrimarySegment = "false"
                isTargetReusedLocationString = "false"
914

915
            result[hostname] += '%s:%d:%s:%s:%d:%d:%s:%s' % (seg.getSegmentDataDirectory(), seg.getSegmentPort(),
S
Shoaib Lari 已提交
916 917
                                                          isPrimarySegment,
                                                          isTargetReusedLocationString,
918
                                                          seg.getSegmentDbId(),
919
                                                          seg.getSegmentContentId(),
920 921
                                                          primaryHostname,
                                                          primarySegmentPort
922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994
            )
        return result

#-----------------------------------------------
class GpVersion(Command):
    def __init__(self,name,gphome,ctxt=LOCAL,remoteHost=None):
        # XXX this should make use of the gphome that was passed
        # in, but this causes problems in some environments and
        # requires further investigation.

        self.gphome=gphome
        #self.cmdStr="%s/bin/postgres --gp-version" % gphome
        self.cmdStr="$GPHOME/bin/postgres --gp-version"
        Command.__init__(self,name,self.cmdStr,ctxt,remoteHost)

    def get_version(self):
        return self.results.stdout.strip()

    @staticmethod
    def local(name,gphome):
        cmd=GpVersion(name,gphome)
        cmd.run(validateAfter=True)
        return cmd.get_version()

#-----------------------------------------------
class GpCatVersion(Command):
    """
    Get the catalog version of the binaries in a given GPHOME
    """
    def __init__(self,name,gphome,ctxt=LOCAL,remoteHost=None):
        # XXX this should make use of the gphome that was passed
        # in, but this causes problems in some environments and
        # requires further investigation.
        self.gphome=gphome
        #cmdStr="%s/bin/postgres --catalog-version" % gphome
        cmdStr="$GPHOME/bin/postgres --catalog-version"
        Command.__init__(self,name,cmdStr,ctxt,remoteHost)

    def get_version(self):
        # Version comes out like this:
        #   "Catalog version number:               201002021"
        # We only want the number
        return self.results.stdout.split(':')[1].strip()

    @staticmethod
    def local(name,gphome):
        cmd=GpCatVersion(name,gphome)
        cmd.run(validateAfter=True)
        return cmd.get_version()

#-----------------------------------------------
class GpCatVersionDirectory(Command):
    """
    Get the catalog version of a given database directory
    """
    def __init__(self,name,directory,ctxt=LOCAL,remoteHost=None):
        cmdStr = "$GPHOME/bin/pg_controldata %s" % directory
        Command.__init__(self,name,cmdStr,ctxt,remoteHost)

    def get_version(self):
        "sift through pg_controldata looking for the catalog version number"
        for key, value in self.results.split_stdout():
            if key == 'Catalog version number':
                return value.strip()

    @staticmethod
    def local(name,directory):
        cmd=GpCatVersionDirectory(name,directory)
        cmd.run(validateAfter=True)
        return cmd.get_version()

#-----------------------------------------------
class GpAddConfigScript(Command):
995
    def __init__(self, name, directorystring, entry, value=None, removeonly=False, ctxt=LOCAL, remoteHost=None):
996 997 998 999 1000 1001 1002 1003 1004 1005
        cmdStr="echo '%s' | $GPHOME/sbin/gpaddconfig.py --entry %s" % (directorystring, entry)
        if value:
            # value will be encoded and unencoded in the script to protect against shell interpretation
            value = base64.urlsafe_b64encode(pickle.dumps(value))
            cmdStr = cmdStr + " --value '" + value + "'"
        if removeonly:
            cmdStr = cmdStr + " --removeonly "

        Command.__init__(self,name,cmdStr,ctxt,remoteHost)

M
Marbin Tan 已提交
1006
#-----------------------------------------------
1007 1008 1009 1010
class GpAppendGucToFile(Command):

    # guc value will come in pickled and base64 encoded

M
Marbin Tan 已提交
1011
    def __init__(self,name,file,guc,value,ctxt=LOCAL,remoteHost=None):
1012 1013 1014 1015 1016
        unpickledText = pickle.loads(base64.urlsafe_b64decode(value))
        finalText = unpickledText.replace('"', '\\\"')
        cmdStr = 'echo "%s=%s" >> %s' %  (guc, finalText, file)
        Command.__init__(self,name,cmdStr,ctxt,remoteHost)

N
Nadeem Ghani 已提交
1017
#-----------------------------------------------
1018
class GpLogFilter(Command):
M
Marbin Tan 已提交
1019
    def __init__(self, name, filename, start=None, end=None, duration=None,
1020
                 case=None, count=None, search_string=None,
M
Marbin Tan 已提交
1021
                 exclude_string=None, search_regex=None, exclude_regex=None,
1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
                 trouble=None, ctxt=LOCAL,remoteHost=None):
        cmdfrags = []
        if start:
            cmdfrags.append('--begin=%s' % start)
        if end:
            cmdfrags.append('--end=%s' % end)
        if duration:
            cmdfrags.append('--duration=%s' % duration)
        if case:
            cmdfrags.append('--case=%s' % case)
        if search_string:
            cmdfrags.append('--find=\'%s\'' % search_string)
        if exclude_string:
            cmdfrags.append('--nofind=\'%s\'' % exclude_string)
        if search_regex:
            cmdfrags.append('--match=\'%s\'' % search_regex)
        if count:
            cmdfrags.append('-n %s' % count)
        if exclude_regex:
            cmdfrags.append('--nomatch=\'%s\'' % exclude_regex)
        if trouble:
            cmdfrags.append('-t')
        cmdfrags.append(filename)
N
Nadeem Ghani 已提交
1045
        self.cmdStr = "$GPHOME/bin/gplogfilter %s" % ' '.join(cmdfrags)
1046
        Command.__init__(self, name, self.cmdStr, ctxt,remoteHost)
N
Nadeem Ghani 已提交
1047 1048

    @staticmethod
M
Marbin Tan 已提交
1049
    def local(name, filename, start=None, end=None, duration=None,
1050
               case=None, count=None, search_string=None,
M
Marbin Tan 已提交
1051
               exclude_string=None, search_regex=None, exclude_regex=None,
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
               trouble=None):
        cmd = GpLogFilter(name, filename, start, end, duration, case, count, search_string,
                          exclude_string, search_regex, exclude_regex, trouble)
        cmd.run(validateAfter=True)
        return "".join(cmd.get_results().stdout).split("\r\n")

#-----------------------------------------------
def distribute_tarball(queue,list,tarball):
        logger.debug("distributeTarBall start")
        for db in list:
            hostname = db.getSegmentHostName()
            datadir = db.getSegmentDataDirectory()
            (head,tail)=os.path.split(datadir)
1065
            scp_cmd=Scp(name="copy master",srcFile=tarball,dstHost=hostname,dstFile=head)
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086
            queue.addCommand(scp_cmd)
        queue.join()
        queue.check_results()
        logger.debug("distributeTarBall finished")




class GpError(Exception): pass

######
def get_gphome():
    gphome=os.getenv('GPHOME',None)
    if not gphome:
        raise GpError('Environment Variable GPHOME not set')
    return gphome


######
def get_masterdatadir():
    master_datadir = os.environ.get('MASTER_DATA_DIRECTORY')
1087
    if not master_datadir:
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
        raise GpError("Environment Variable MASTER_DATA_DIRECTORY not set!")
    return master_datadir

######
def get_masterport(datadir):
    return pgconf.readfile(os.path.join(datadir, 'postgresql.conf')).int('port')


######
def check_permissions(username):
    logger.debug("--Checking that current user can use GP binaries")
    chk_gpdb_id(username)






#=-=-=-=-=-=-=-=-=-= Bash Migration Helper Functions =-=-=-=-=-=-=-=-

1108
def start_standbymaster(host, datadir, port, era=None,
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
                        wrapper=None, wrapper_args=None):
    logger.info("Starting standby master")

    logger.info("Checking if standby master is running on host: %s  in directory: %s" % (host,datadir))
    cmd = Command("recovery_startup",
                  ("python -c "
                   "'from gppylib.commands.gp import recovery_startup; "
                   """recovery_startup("{0}", "{1}")'""").format(
                       datadir, port),
                  ctxt=REMOTE, remoteHost=host)
    cmd.run()
    res = cmd.get_results().stderr

    if res:
        logger.warning("Unable to cleanup previously started standby: '%s'" % res)

    cmd = GpStandbyStart.remote('start standby master',
1126
                                host, datadir, port, era=era,
1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
                                wrapper=wrapper, wrapper_args=wrapper_args)
    logger.debug("Starting standby: %s" % cmd )

    logger.debug("Starting standby master results: %s" % cmd.get_results() )

    if cmd.get_results().rc != 0:
        logger.warning("Could not start standby master: %s" % cmd)
        return False

    # Wait for the standby to start recovery.  Ideally this means the
D
Daniel Gustafsson 已提交
1137
    # standby connection is recognized by the primary, but locally in this
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
    # function it is better to work with only standby.  If recovery has
    # started, this means now postmaster is responsive to signals, which
    # allows shutdown etc.  If we exit earlier, there is a big chance
    # a shutdown message from other process is missed.
    for i in xrange(60):
        # Fetch it every time, as postmaster might not have been up yet for
        # the first few cycles, which we have seen when trying wrapper
        # shell script.
        pid = getPostmasterPID(host, datadir)
        cmd = Command("get pids",
                      ("python -c "
                       "'from gppylib.commands import unix; "
                       "print unix.getDescendentProcesses({0})'".format(pid)),
                      ctxt=REMOTE, remoteHost=host)
        cmd.run()
        logger.debug(str(cmd))
        result = cmd.get_results()
        logger.debug(result)
        # We want more than postmaster and logger processes.
        if result.rc == 0 and len(result.stdout.split(',')) > 2:
            return True
        time.sleep(1)

    logger.warning("Could not start standby master")
    return False

def get_pid_from_remotehost(host, datadir):
    cmd = Command(name = 'get the pid from postmaster file',
M
Marbin Tan 已提交
1166 1167
                  cmdStr = 'head -1 %s/postmaster.pid' % datadir,
                  ctxt=REMOTE, remoteHost = host)
1168
    cmd.run()
M
Marbin Tan 已提交
1169
    pid = None
1170 1171 1172 1173 1174 1175
    if cmd.get_results().rc == 0 and cmd.get_results().stdout.strip():
        pid = int(cmd.get_results().stdout.strip())
    return pid

def is_pid_postmaster(datadir, pid, remoteHost=None):
    """
D
Daniel Gustafsson 已提交
1176
    This function returns true on any uncertainty: if it cannot execute pgrep, pwdx or just connect to the standby host
1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
    it will return true
    """
    def validate_command (commandName, datadir, ctxt, remoteHost):
        cmd = Command ('Check %s availability' % commandName, commandName, ctxt=ctxt, remoteHost=remoteHost)
        cmd.run()
        if cmd.get_results().rc == COMMAND_NOT_FOUND:
            if not remoteHost is None:
                logger.warning('command "%s" is not found on host %s. cannot check postmaster status, assuming it is running', commandName, remoteHost)
            else:
                logger.warning('command "%s" is not found. cannot check postmaster status, assuming it is running', commandName)
            return False
        return True

    if remoteHost is not None:
        ctxt = REMOTE
    else:
        ctxt = LOCAL
M
Marbin Tan 已提交
1194

1195 1196 1197
    is_postmaster = True
    if (validate_command ('pgrep', datadir, ctxt, remoteHost) and
            validate_command ('pwdx', datadir, ctxt, remoteHost)):
1198
        cmdStr = 'pgrep postgres | xargs -I{} pwdx {} | grep "%s" | grep "^%s:" | cat' % (datadir, pid)
1199 1200 1201 1202 1203 1204 1205 1206
        cmd = Command("search for postmaster process", cmdStr, ctxt=ctxt, remoteHost=remoteHost)
        res = None
        try:
            cmd.run(validateAfter=True)
            res = cmd.get_results()
            if not res.stdout.strip():
                is_postmaster = False
            else:
M
Marbin Tan 已提交
1207
                logger.info(res.stdout.strip())
1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389
        except Exception as e:
            if not remoteHost is None:
                logger.warning('failed to get the status of postmaster %s on %s. assuming that postmaster is running' % (datadir, remoteHost))
            else:
                logger.warning('failed to get the status of postmaster %s. assuming that postmaster is running' % (datadir))

    return is_postmaster

######
def recovery_startup(datadir, port=None):
    """ investigate a db that may still be running """

    pid=read_postmaster_pidfile(datadir)
    if check_pid(pid) and is_pid_postmaster(datadir, pid):
        info_str="found postmaster with pid: %d for datadir: %s still running" % (pid,datadir)
        logger.info(info_str)

        logger.info("attempting to shutdown db with datadir: %s" % datadir )
        cmd=SegmentStop('db shutdown' , datadir,mode='fast')
        cmd.run()

        if check_pid(pid) and is_pid_postmaster(datadir, pid):
            info_str="unable to stop postmaster with pid: %d for datadir: %s still running" % (pid,datadir)
            logger.info(info_str)
            return info_str
        else:
            logger.info("shutdown of db successful with datadir: %s" % datadir)
            return None
    else:
        # If we get this far it means we don't have a pid and need to do some
        # cleanup.  Use port number if supplied.  postgresql.conf may be
        # bogus as we pass port number anyway instead of postgresql.conf
        # default value.
        if port is None:
            pgconf_dict = pgconf.readfile(datadir + "/postgresql.conf")
            port = pgconf_dict.int('port')

        lockfile="/tmp/.s.PGSQL.%s" % port
        tmpfile_exists = os.path.exists(lockfile)


        logger.info("No db instance process, entering recovery startup mode")

        if tmpfile_exists:
            logger.info("Clearing db instance lock files")
            os.remove(lockfile)

        postmaster_pid_file = "%s/postmaster.pid" % datadir
        if os.path.exists(postmaster_pid_file):
            logger.info("Clearing db instance pid file")
            os.remove("%s/postmaster.pid" % datadir)

        return None


# these match names from gp_bash_functions.sh
def chk_gpdb_id(username):
    path="%s/bin/initdb" % GPHOME
    if not os.access(path,os.X_OK):
        raise GpError("File permission mismatch.  The current user %s does not have sufficient"
                      " privileges to run the Greenplum binaries and management utilities." % username )


def chk_local_db_running(datadir, port):
    """Perform a few checks to see if the db is running.  We 1st look at:

       1) /tmp/.s.PGSQL.<PORT> and /tmp/.s.PGSQL.<PORT>.lock
       2) DATADIR/postmaster.pid
       3) netstat

       Returns tuple in format (postmaster_pid_file_exists, tmpfile_exists, lockfile_exists, port_active, postmaster_pid)

       postmaster_pid value is 0 if postmaster_pid_exists is False  Note that this is the PID from the postmaster.pid file

    """

    # determine if postmaster.pid is there, grab pid from it
    postmaster_pid_exists = True
    f = None
    try:
        f = open(datadir + "/postmaster.pid")
    except IOError:
        postmaster_pid_exists = False

    pid_value = 0
    if postmaster_pid_exists:
        try:
            for line in f:
                pid_value = int(line) # grab first line only
                break
        finally:
            f.close()

    cmd=FileDirExists('check for /tmp/.s.PGSQL file file', "/tmp/.s.PGSQL.%d" % port)
    cmd.run(validateAfter=True)
    tmpfile_exists = cmd.filedir_exists()

    cmd=FileDirExists('check for lock file', get_lockfile_name(port))
    cmd.run(validateAfter=True)
    lockfile_exists = cmd.filedir_exists()

    netstat_port_active = PgPortIsActive.local('check netstat for postmaster port',"/tmp/.s.PGSQL.%d" % port, port)

    logger.debug("postmaster_pid_exists: %s tmpfile_exists: %s lockfile_exists: %s netstat port: %s  pid: %s" %\
                (postmaster_pid_exists, tmpfile_exists, lockfile_exists, netstat_port_active, pid_value))

    return (postmaster_pid_exists, tmpfile_exists, lockfile_exists, netstat_port_active, pid_value)

def get_lockfile_name(port):
    return "/tmp/.s.PGSQL.%d.lock" % port


def get_local_db_mode(master_data_dir):
    """ Gets the mode Greenplum is running in.
        Possible return values are:
            'NORMAL'
            'RESTRICTED'
            'UTILITY'
    """
    mode = 'NORMAL'

    if not os.path.exists(master_data_dir + '/postmaster.pid'):
        raise Exception('Greenplum database appears to be stopped')

    try:
        fp = open(master_data_dir + '/postmaster.opts', 'r')
        optline = fp.readline()
        if optline.find('superuser_reserved_connections') > 0:
            mode = 'RESTRICTED'
        elif optline.find('gp_role=utility') > 0:
            mode = 'UTILITY'
    except OSError:
        raise Exception('Failed to open %s.  Is Greenplum Database running?' % master_data_dir + '/postmaster.opts')
    except IOError:
        raise Exception('Failed to read options from %s' % master_data_dir + '/postmaster.opts')
    finally:
        if fp: fp.close()

    return mode

######
def read_postmaster_pidfile(datadir, host=None):
    if host:
        cmdStr ="""python -c 'from {module} import {func}; print {func}("{args}")'""".format(module=sys.modules[__name__].__name__,
                                                                                             func='read_postmaster_pidfile',
                                                                                             args=datadir)
        cmd = Command(name='run this method remotely', cmdStr=cmdStr, ctxt=REMOTE, remoteHost=host)
        cmd.run(validateAfter=True)
        return int(cmd.get_results().stdout.strip())
    pid=0
    f = None
    try:
        f = open(datadir + '/postmaster.pid')
        pid = int(f.readline().strip())
    except Exception:
        pass
    finally:
        if f: f.close()
    return pid


def createTempDirectoryName(masterDataDirectory, tempDirPrefix):
    return '%s/%s_%s_%d' % (os.sep.join(os.path.normpath(masterDataDirectory).split(os.sep)[:-1]),
                                tempDirPrefix,
                                datetime.datetime.now().strftime('%m%d%Y'),
                                os.getpid())

#-------------------------------------------------------------------------
class GpRecoverSeg(Command):
   """
   This command will execute the gprecoverseg utility
   """

   def __init__(self, name, options = "", ctxt = LOCAL, remoteHost = None):
       self.name = name
       self.options = options
       self.ctxt = ctxt
       self.remoteHost = remoteHost

       cmdStr = "$GPHOME/bin/gprecoverseg %s" % (options)
       Command.__init__(self,name,cmdStr,ctxt,remoteHost)

L
Larry Hamel 已提交
1390
class GpReadConfig(Command):
J
Jamie McAtamney 已提交
1391
    def __init__(self, name, seg, guc_name):
L
Larry Hamel 已提交
1392 1393 1394
        self.seg_db_id = seg.getSegmentDbId()
        self.seg_content_id = seg.getSegmentContentId()
        self.guc_name = guc_name
1395
        self.role = seg.getSegmentRole()
L
Larry Hamel 已提交
1396 1397 1398
        cat_path = findCmdInPath('cat')

        cmdStr = "%s %s/postgresql.conf" % (cat_path, seg.getSegmentDataDirectory())
J
Jamie McAtamney 已提交
1399 1400 1401 1402 1403
        ctxt = LOCAL
        remote_host = None
        if seg.hostname != socket.gethostname():
            ctxt = REMOTE
            remote_host = seg.hostname
L
Larry Hamel 已提交
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423
        Command.__init__(self, name, cmdStr, ctxt, remote_host)

    def get_guc_value(self):
        std_out = self.get_results().stdout
        std_out = std_out.split('\n')

        VALUE_PATTERN = re.compile(".*=(.*)")

        GUC_PATTERN = re.compile("^[\s]*" + self.guc_name + "[ \t]*=")

        value = None
        key_lines = [line for line in std_out if GUC_PATTERN.match(line)]
        if key_lines:
            value = VALUE_PATTERN.match(key_lines[-1]).group(1)
            value = value.split('#')[0].strip()

        return value

    def get_seg_content_id(self):
        return self.seg_content_id
1424

1425 1426 1427 1428 1429 1430
    def get_seg_role(self):
        return self.role

    def get_seg_dbid(self):
        return self.seg_db_id

1431 1432 1433 1434 1435

if __name__ == '__main__':

    import doctest
    doctest.testmod()