def test_getandplay(self):
util.print_frame()
test_limit_mb = 10
server = self.cluster['servers'][0]
redis = telnetlib.Telnet(server['ip'], server['redis_port'])
util.log("Insert large key about 100MB")
self.insertLargeKey(redis, "test_key")
# Test getandplay
start_time = time.time()
util.log("Start getandplay, start ts:%d" % start_time)
cmd = "./cluster-util --getandplay %s %d %s %d 0-8191 30000 %d" % (server['ip'], self.getdump_proxy_port,
server['ip'], self.playdump_proxy_port, test_limit_mb)
proc = util.exec_proc_async(util.cluster_util_dir(0), cmd, True, None, subprocess.PIPE, None)
monitor_file1 = "%s/%s" % (util.cluster_util_dir(0), self.getdump_proxy_log)
monitor_file2 = "%s/%s" % (util.cluster_util_dir(0), self.playdump_proxy_log)
self.monitor_filesize_diff(proc, test_limit_mb, monitor_file1, monitor_file2)
ret = proc.wait()
self.assertEqual(0, ret)
elapse_time = time.time() - start_time
util.log("End getandplay, elapsed:%d" % elapse_time)
dump_file_size = os.path.getsize("%s/%s" % (util.cluster_util_dir(0), self.getdump_proxy_log))
play_file_size = os.path.getsize("%s/%s" % (util.cluster_util_dir(0), self.playdump_proxy_log))
util.log("Dump File Size:%d, Play File Size:%d, elapsed:%d, limit:%dMB/s, actual:%dMB/s"
% (dump_file_size, play_file_size, elapse_time, test_limit_mb,
(dump_file_size + play_file_size) / elapse_time / 1024 / 1024))
self.assertTrue((dump_file_size + play_file_size) / (10 * 1024 * 1024) < elapse_time)
def test_catdatadel(self):
util.print_frame()
test_limit_mb = 10
# Make data
server = self.cluster['servers'][0]
redis = telnetlib.Telnet(server['ip'], server['redis_port'])
util.log("Insert large key about 100MB")
self.insertLargeKey(redis, "test_key")
# Get dump
util.log("Start getdump, start ts:%d" % time.time())
cmd = "./cluster-util --getdump %s %d getdump.rdb 0-8191 %d" % (server['ip'], server['redis_port'], test_limit_mb)
proc = util.exec_proc_async(util.cluster_util_dir(0), cmd, True, None, subprocess.PIPE, None)
ret = proc.wait()
self.assertEqual(0, ret)
# Test --catdata sub command
util.log("Start --catdata, start ts:%d" % time.time())
cmd = "./cluster-util --catdata getdump.rdb > catdata.resp"
proc = util.exec_proc_async(util.cluster_util_dir(0), cmd, True, None, subprocess.PIPE, None)
ret = proc.wait()
self.assertEqual(0, ret)
# test --deldata sub command
util.log("Start --deldata, start ts:%d" % time.time())
cmd = "./cluster-util --deldata getdump.rdb > deldata.resp"
proc = util.exec_proc_async(util.cluster_util_dir(0), cmd, True, None, subprocess.PIPE, None)
ret = proc.wait()
self.assertEqual(0, ret)
def setup_proxy_server(self, in_port, out_port, in_log, out_log, fifo_name):
try:
os.remove("%s/%s" % (util.cluster_util_dir(0), fifo_name))
except Exception as e:
pass
fifo_cmd = "mkfifo %s" % fifo_name
fifo_proc = util.exec_proc_async(util.cluster_util_dir(0), fifo_cmd, True, None, subprocess.PIPE, None)
ret = fifo_proc.wait()
self.assertEqual(0, ret)
proxy_cmd = "nc -l %d < %s | tee %s | nc 127.0.0.1 %d | tee %s > %s" % (
in_port, fifo_name, in_log, out_port, out_log, fifo_name)
return util.exec_proc_async(util.cluster_util_dir(0), proxy_cmd, True, None, subprocess.PIPE, None)
def test_getdump_and_playdump(self):
util.print_frame()
self.setup_proxies()
test_limit_mb = 10
server = self.cluster['servers'][0]
redis = telnetlib.Telnet(server['ip'], server['redis_port'])
util.log("Insert large key about 100MB")
self.insertLargeKey(redis, "test_key")
# Test getdump
start_time = time.time()
util.log("Start getdump, start ts:%d" % start_time)
cmd = "./cluster-util --getdump %s %d getdump.rdb 0-8191 %d" % (server['ip'], self.getdump_proxy_port, test_limit_mb)
proc = util.exec_proc_async(util.cluster_util_dir(0), cmd, True, None, subprocess.PIPE, None)
monitor_file = "%s/%s" % (util.cluster_util_dir(0), self.getdump_proxy_log)
self.monitor_filesize_diff(proc, test_limit_mb, monitor_file)
ret = proc.wait()
self.assertEqual(0, ret)
elapse_time = time.time() - start_time
util.log("End getdump, elapsed:%d" % elapse_time)
file_size = os.path.getsize("%s/%s" % (util.cluster_util_dir(0), self.getdump_proxy_log))
util.log("File Size:%d, elapsed:%d, limit:%dMB/s, actual:%dMB/s" % (file_size,
elapse_time, test_limit_mb, file_size / elapse_time / 1024 / 1024))
self.assertTrue(file_size / (10 * 1024 * 1024) < elapse_time)
# Test playdump
start_time = time.time()
util.log("Start playdump, start ts:%d" % start_time)
cmd = "./cluster-util --playdump getdump.rdb %s %d 30000 %d" % (server['ip'], self.playdump_proxy_port, test_limit_mb)
proc = util.exec_proc_async(util.cluster_util_dir(0), cmd, True, None, subprocess.PIPE, None)
monitor_file = "%s/%s" % (util.cluster_util_dir(0), self.playdump_proxy_log)
self.monitor_filesize_diff(proc, test_limit_mb, monitor_file)
ret = proc.wait()
self.assertEqual(0, ret)
elapse_time = time.time() - start_time
util.log("End playdump, elapsed:%d" % elapse_time)
file_size = os.path.getsize("%s/%s" % (util.cluster_util_dir(0), self.playdump_proxy_log))
util.log("File Size:%d, elapsed:%d, limit:%dMB/s, actual:%dMB/s" % (file_size,
elapse_time, test_limit_mb, file_size / elapse_time / 1024 / 1024))
self.assertTrue(file_size / (10 * 1024 * 1024) < elapse_time)
def numOpenFds(self, pid):
p = util.exec_proc_async(util.cluster_util_dir(0), 'ls /proc/%s/fd | wc -l' % pid,
True, None, subprocess.PIPE, None)
ret = p.wait()
return p.stdout.read()[:-1]
def test_migration_with_expire_command(self):
util.print_frame()
util.log("start load_generator")
load_gen_thrd_list = {}
for i in range(1):
ip, port = util.get_rand_gateway(self.cluster)
load_gen_thrd_list[i] = load_generator.LoadGenerator(i, ip, port)
load_gen_thrd_list[i].start()
time.sleep(5) # generate load for 5 sec
tps = 20000
src_pg_id = 0
dst_pg_id = 1
leader_cm = self.cluster['servers'][0]
src_master = util.get_server_by_role_and_pg(self.cluster['servers'], 'master', src_pg_id)
dst_master = util.get_server_by_role_and_pg(self.cluster['servers'], 'master', dst_pg_id)
smr = smr_mgmt.SMR(src_master['id'])
ret = smr.connect(src_master['ip'], src_master['smr_mgmt_port'])
if ret != 0:
util.log('failed to connect to smr(source master)')
return False
src_redis = redis_mgmt.Redis(src_master['id'])
ret = src_redis.connect(src_master['ip'], src_master['redis_port'] )
self.assertEquals( ret, 0, 'failed to connect to redis' )
dst_redis = redis_mgmt.Redis(dst_master['id'])
ret = dst_redis.connect(dst_master['ip'], dst_master['redis_port'] )
self.assertEquals( ret, 0, 'failed to connect to redis' )
ts = time.time()
self.setExpireKey(src_redis, 'beforeCheckpoint~beforeCheckpoint:expired', 10)
self.setExpireKey(src_redis, 'beforeCheckpoint~beforeCheckpoint:persist', 20)
self.setExpireS3Key(src_redis, 'S3:beforeCheckpoint~beforeCheckpoint:expired', 10)
self.setExpireS3Key(src_redis, 'S3:beforeCheckpoint~beforeCheckpoint:persist', 20)
self.setExpireS3Key(src_redis, 'S3:PermanentKey', 0)
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis, 'beforeCheckpoint~beforeCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis, 'beforeCheckpoint~beforeCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis, 'S3:beforeCheckpoint~beforeCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis, 'S3:beforeCheckpoint~beforeCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
util.log(">>> migrate test with expire command start(%s), ts:%d" % (time.asctime(), ts))
ts = time.time()
self.setExpireKey(src_redis, 'beforeCheckpoint~afterCheckpoint:expired', 10)
self.setExpireKey(src_redis, 'beforeCheckpoint~afterCheckpoint:persist', 20)
self.setExpireS3Key(src_redis, 'S3:beforeCheckpoint~afterCheckpoint:expired', 10)
self.setExpireS3Key(src_redis, 'S3:beforeCheckpoint~afterCheckpoint:persist', 20)
# notify dst_redis of migration start
util.log(">>> notify dst_redis of migration start (%s)" % time.asctime())
cmd = 'migconf migstart %d-%d\r\n' % (0, 8191)
dst_redis.write(cmd)
res = dst_redis.read_until('\r\n')
self.assertEquals( res, '+OK\r\n' )
# remote partial checkpoint
util.log(">>> start remote checkpoint and load (%s)" % time.asctime())
cmd = "./cluster-util --getandplay %s %d %s %d %d-%d %d" % (
src_master['ip'], src_master['redis_port'],
dst_master['ip'], dst_master['redis_port'],
0, 8191, tps)
p = util.exec_proc_async(util.cluster_util_dir(src_master['id']), cmd, True, None, subprocess.PIPE, None)
ret = p.wait()
for line in p.stdout:
if line.find("Checkpoint Sequence Number:") != -1:
util.log("seqnumber : " + line[line.rfind(":")+1:])
seq = int(line[line.rfind(":")+1:])
util.log(">>>" + str(line.rstrip()))
self.assertEqual(0, ret)
util.log(">>> end remote checkpoint and load (%s)" % time.asctime())
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis, 'beforeCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis, 'beforeCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis, 'S3:beforeCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis, 'S3:beforeCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
# bgsave for testing later about recovery during migration
util.log(">>> bgsave for testing later about recovery during migration (%s)" % time.asctime())
cmd = 'bgsave\r\n'
dst_redis.write(cmd)
res = dst_redis.read_until('\r\n')
self.assertEquals( res, '+Background saving started\r\n' )
ts = time.time()
self.setExpireKey(src_redis, 'afterCheckpoint~afterCheckpoint:expired', 10)
self.setExpireKey(src_redis, 'afterCheckpoint~afterCheckpoint:persist', 20)
self.setExpireS3Key(src_redis, 'S3:afterCheckpoint~afterCheckpoint:expired', 10)
self.setExpireS3Key(src_redis, 'S3:afterCheckpoint~afterCheckpoint:persist', 20)
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis, 'afterCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis, 'afterCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis, 'S3:afterCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis, 'S3:afterCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(src_redis, 'afterCheckpoint~duringCatchup:expired', 10)
self.setExpireKey(src_redis, 'afterCheckpoint~duringCatchup:persist', 100)
self.setExpireS3Key(src_redis, 'S3:afterCheckpoint~duringCatchup:expired', 10)
self.setExpireS3Key(src_redis, 'S3:afterCheckpoint~duringCatchup:persist', 100)
# remote catchup (smr log migration)
util.log(">>> start remote catchup (%s)" % time.asctime())
dst_host = dst_master['ip']
dst_smr_port = dst_master['smr_base_port']
rle = '1 8192'
num_part = 8192
smr.write('migrate start %s %d %d %d %d %s\r\n' % (dst_host, dst_smr_port,
seq, tps, num_part, rle))
response = smr.read_until('\r\n')
if response[:3] != '+OK':
util.log('failed to execute migrate start command, response:%s' % response)
return False
while True:
smr.write('migrate info\r\n')
response = smr.read_until('\r\n')
seqs = response.split()
logseq = int(seqs[1].split(':')[1])
mig = int(seqs[2].split(':')[1])
util.log('migrate info: %s' % response)
if (logseq-mig < 500000):
util.log('Remote catchup almost done. try mig2pc')
break
time.sleep(1)
util.log(">>> sleep until 90 sec pass")
self.assertFalse(time.time() - ts >= 90)
time.sleep(90 - (time.time() - ts))
res = self.persistKey(src_redis, 'afterCheckpoint~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis, 'afterCheckpoint~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis, 'S3:afterCheckpoint~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis, 'S3:afterCheckpoint~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(src_redis, 'duringCatchup~duringCatchup:expired', 10)
self.setExpireKey(src_redis, 'duringCatchup~duringCatchup:persist', 20)
self.setExpireS3Key(src_redis, 'S3:duringCatchup~duringCatchup:expired', 10)
self.setExpireS3Key(src_redis, 'S3:duringCatchup~duringCatchup:persist', 20)
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis, 'duringCatchup~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis, 'duringCatchup~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis, 'S3:duringCatchup~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis, 'S3:duringCatchup~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(src_redis, 'duringCatchup~afterMig2pc:expired', 10)
self.setExpireKey(src_redis, 'duringCatchup~afterMig2pc:persist', 20)
self.setExpireS3Key(src_redis, 'S3:duringCatchup~afterMig2pc:expired', 10)
self.setExpireS3Key(src_redis, 'S3:duringCatchup~afterMig2pc:persist', 20)
util.log(">>> remote catchup phase almost done (%s)" % time.asctime())
# mig2pc
util.log(">>> start mig2pc (%s)" % time.asctime())
cmd = 'mig2pc %s %d %d %d %d' % (self.cluster['cluster_name'], src_pg_id, dst_pg_id,
0, 8191)
result = util.cm_command(leader_cm['ip'], leader_cm['cm_port'], cmd)
util.log('mig2pc result : ' + result)
if not result.startswith('{"state":"success","msg":"+OK"}\r\n'):
util.log('failed to execute mig2pc command, result:%s' % result)
return False
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(dst_redis, 'duringCatchup~afterMig2pc:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(dst_redis, 'duringCatchup~afterMig2pc:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(dst_redis, 'S3:duringCatchup~afterMig2pc:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(dst_redis, 'S3:duringCatchup~afterMig2pc:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(dst_redis, 'afterMig2pc~migrateEnd:expired', 10)
self.setExpireKey(dst_redis, 'afterMig2pc~migrateEnd:persist', 20)
self.setExpireS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:expired', 10)
self.setExpireS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:persist', 20)
# finish migration
smr.write('migrate interrupt\r\n')
response = smr.read_until('\r\n')
util.log('migrate interrupt: %s' % response)
smr.disconnect()
# notify dst_redis of migration end
util.log(">>> notify dst_redis of migration end (%s)" % time.asctime())
cmd = 'migconf migend\r\n'
dst_redis.write(cmd)
res = dst_redis.read_until('\r\n')
self.assertEquals( res, '+OK\r\n' )
cmd = 'migconf clearstart %d-%d\r\n' % (0, 8191)
src_redis.write(cmd)
res = src_redis.read_until('\r\n')
self.assertEquals( res, '+OK\r\n' )
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(dst_redis, 'afterMig2pc~migrateEnd:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(dst_redis, 'afterMig2pc~migrateEnd:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
self.assertTrue(self.isExist(dst_redis, 'beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis, 'beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis, 'beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis, 'beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:persist'))
self.assertFalse(self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~duringCatchup:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~duringCatchup:expired'))
self.assertTrue(self.isExist(dst_redis, 'duringCatchup~duringCatchup:persist'))
self.assertFalse(self.isExist(dst_redis, 'duringCatchup~duringCatchup:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:duringCatchup~duringCatchup:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:duringCatchup~duringCatchup:expired'))
self.assertTrue(self.isExist(dst_redis, 'duringCatchup~afterMig2pc:persist'))
self.assertFalse(self.isExist(dst_redis, 'duringCatchup~afterMig2pc:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:expired'))
self.assertTrue(self.isExist(dst_redis, 'afterMig2pc~migrateEnd:persist'))
self.assertFalse(self.isExist(dst_redis, 'afterMig2pc~migrateEnd:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:expired'))
# remote partial checkpoint
util.log(">>> start rangedel (%s)" % time.asctime())
cmd = "./cluster-util --rangedel %s %d %d-%d %d" % (
src_master['ip'], src_master['redis_port'],
0, 8191, tps)
p = util.exec_proc_async(util.cluster_util_dir(src_master['id']), cmd, True, None, subprocess.PIPE, None)
ret = p.wait()
for line in p.stdout:
util.log(">>>" + str(line.rstrip()))
cmd = 'migconf clearend\r\n'
src_redis.write(cmd)
res = src_redis.read_until('\r\n')
self.assertEqual(res, '+OK\r\n')
time.sleep(5) # generate load for 5 sec
# check consistency of load_generator
for i in range(len(load_gen_thrd_list)):
load_gen_thrd_list[i].quit()
for i in range(len(load_gen_thrd_list)):
load_gen_thrd_list[i].join()
self.assertTrue(load_gen_thrd_list[i].isConsistent(), 'Inconsistent after migration')
# kill dst_redis and recover from bgsave
util.log(">>> kill dst_redis and recover from bgsave (%s)" % time.asctime())
dst_redis.disconnect()
ret = testbase.request_to_shutdown_redis(dst_master)
self.assertEquals( ret, 0, 'failed to shutdown redis' )
ret = testbase.request_to_shutdown_smr(dst_master)
self.assertEquals(ret, 0, 'failed to shutdown smr')
time.sleep(5)
testbase.request_to_start_smr(dst_master)
self.assertEqual( ret, 0, 'failed to start smr, server:%d' % dst_master['id'] )
ret = testbase.request_to_start_redis(dst_master)
self.assertEqual( ret, 0, 'failed to start redis, server:%d' % dst_master['id'] )
ret = testbase.wait_until_finished_to_set_up_role(dst_master)
self.assertEquals( ret, 0, 'failed to role change. server:%d' % (dst_master['id']) )
dst_redis = redis_mgmt.Redis(dst_master['id'])
ret = dst_redis.connect(dst_master['ip'], dst_master['redis_port'] )
self.assertEquals( ret, 0, 'failed to connect to redis' )
self.assertTrue(self.isExist(dst_redis, 'beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis, 'beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis, 'beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis, 'beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:persist'))
self.assertFalse(self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~duringCatchup:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:afterCheckpoint~duringCatchup:expired'))
self.assertTrue(self.isExist(dst_redis, 'duringCatchup~duringCatchup:persist'))
self.assertFalse(self.isExist(dst_redis, 'duringCatchup~duringCatchup:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:duringCatchup~duringCatchup:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:duringCatchup~duringCatchup:expired'))
self.assertTrue(self.isExist(dst_redis, 'duringCatchup~afterMig2pc:persist'))
self.assertFalse(self.isExist(dst_redis, 'duringCatchup~afterMig2pc:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:expired'))
self.assertTrue(self.isExist(dst_redis, 'afterMig2pc~migrateEnd:persist'))
self.assertFalse(self.isExist(dst_redis, 'afterMig2pc~migrateEnd:expired'))
self.assertTrue(self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:persist'))
self.assertFalse(self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:expired'))
self.getS3TTL(dst_redis, 'S3:PermanentKey')
# kill dst_slave redis and recover without dump file
util.log(">>> kill dst_redis and recover without dump file (%s)" % time.asctime())
dst_slave = util.get_server_by_role_and_pg(self.cluster['servers'], 'slave', dst_pg_id)
ret = testbase.request_to_shutdown_redis(dst_slave)
self.assertEquals( ret, 0, 'failed to shutdown redis' )
ret = testbase.request_to_shutdown_smr(dst_slave)
self.assertEquals(ret, 0, 'failed to shutdown smr')
time.sleep(5)
testbase.request_to_start_smr(dst_slave)
self.assertEqual( ret, 0, 'failed to start smr, server:%d' % dst_slave['id'] )
ret = testbase.request_to_start_redis(dst_slave)
self.assertEqual( ret, 0, 'failed to start redis, server:%d' % dst_slave['id'] )
ret = testbase.wait_until_finished_to_set_up_role(dst_slave)
self.assertEquals( ret, 0, 'failed to role change. server:%d' % (dst_slave['id']) )
dst_redis_slave = redis_mgmt.Redis(dst_slave['id'])
ret = dst_redis_slave.connect(dst_slave['ip'], dst_slave['redis_port'] )
self.assertEquals( ret, 0, 'failed to connect to redis' )
self.assertTrue(self.isExist(dst_redis_slave, 'beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis_slave, 'beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis_slave, 'afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(self.isExist(dst_redis_slave, 'afterCheckpoint~duringCatchup:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'afterCheckpoint~duringCatchup:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:afterCheckpoint~duringCatchup:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:afterCheckpoint~duringCatchup:expired'))
self.assertTrue(self.isExist(dst_redis_slave, 'duringCatchup~duringCatchup:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'duringCatchup~duringCatchup:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:duringCatchup~duringCatchup:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:duringCatchup~duringCatchup:expired'))
self.assertTrue(self.isExist(dst_redis_slave, 'duringCatchup~afterMig2pc:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'duringCatchup~afterMig2pc:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:duringCatchup~afterMig2pc:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:duringCatchup~afterMig2pc:expired'))
self.assertTrue(self.isExist(dst_redis_slave, 'afterMig2pc~migrateEnd:persist'))
self.assertFalse(self.isExist(dst_redis_slave, 'afterMig2pc~migrateEnd:expired'))
self.assertTrue(self.isS3Exist(dst_redis_slave, 'S3:afterMig2pc~migrateEnd:persist'))
self.assertFalse(self.isS3Exist(dst_redis_slave, 'S3:afterMig2pc~migrateEnd:expired'))
self.getS3TTL(dst_redis_slave, 'S3:PermanentKey')
# Go back to initial configuration
self.assertTrue(util.migration(self.cluster, dst_pg_id, src_pg_id, 0, 8191, 40000),
'failed to rollback migration')
def release_proxy_server(self, proc, fifo_name):
proc.terminate()
try:
os.remove("%s/%s" % (util.cluster_util_dir(0), fifo_name))
except Exception as e:
pass
def test_migration_with_expire_command(self):
util.print_frame()
util.log("start load_generator")
load_gen_thrd_list = {}
for i in range(1):
ip, port = util.get_rand_gateway(self.cluster)
load_gen_thrd_list[i] = load_generator.LoadGenerator(i, ip, port)
load_gen_thrd_list[i].start()
time.sleep(5) # generate load for 5 sec
tps = 20000
src_pg_id = 0
dst_pg_id = 1
leader_cm = self.cluster['servers'][0]
src_master = util.get_server_by_role_and_pg(self.cluster['servers'],
'master', src_pg_id)
dst_master = util.get_server_by_role_and_pg(self.cluster['servers'],
'master', dst_pg_id)
smr = smr_mgmt.SMR(src_master['id'])
ret = smr.connect(src_master['ip'], src_master['smr_mgmt_port'])
if ret != 0:
util.log('failed to connect to smr(source master)')
return False
src_redis = redis_mgmt.Redis(src_master['id'])
ret = src_redis.connect(src_master['ip'], src_master['redis_port'])
self.assertEquals(ret, 0, 'failed to connect to redis')
dst_redis = redis_mgmt.Redis(dst_master['id'])
ret = dst_redis.connect(dst_master['ip'], dst_master['redis_port'])
self.assertEquals(ret, 0, 'failed to connect to redis')
ts = time.time()
self.setExpireKey(src_redis,
'beforeCheckpoint~beforeCheckpoint:expired', 10)
self.setExpireKey(src_redis,
'beforeCheckpoint~beforeCheckpoint:persist', 20)
self.setExpireS3Key(src_redis,
'S3:beforeCheckpoint~beforeCheckpoint:expired', 10)
self.setExpireS3Key(src_redis,
'S3:beforeCheckpoint~beforeCheckpoint:persist', 20)
self.setExpireS3Key(src_redis, 'S3:PermanentKey', 0)
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis,
'beforeCheckpoint~beforeCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis,
'beforeCheckpoint~beforeCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(
src_redis, 'S3:beforeCheckpoint~beforeCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(
src_redis, 'S3:beforeCheckpoint~beforeCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
util.log(">>> migrate test with expire command start(%s), ts:%d" %
(time.asctime(), ts))
ts = time.time()
self.setExpireKey(src_redis,
'beforeCheckpoint~afterCheckpoint:expired', 10)
self.setExpireKey(src_redis,
'beforeCheckpoint~afterCheckpoint:persist', 20)
self.setExpireS3Key(src_redis,
'S3:beforeCheckpoint~afterCheckpoint:expired', 10)
self.setExpireS3Key(src_redis,
'S3:beforeCheckpoint~afterCheckpoint:persist', 20)
# notify dst_redis of migration start
util.log(">>> notify dst_redis of migration start (%s)" %
time.asctime())
cmd = 'migconf migstart %d-%d\r\n' % (0, 8191)
dst_redis.write(cmd)
res = dst_redis.read_until('\r\n')
self.assertEquals(res, '+OK\r\n')
# remote partial checkpoint
util.log(">>> start remote checkpoint and load (%s)" % time.asctime())
cmd = "./cluster-util --getandplay %s %d %s %d %d-%d %d" % (
src_master['ip'], src_master['redis_port'], dst_master['ip'],
dst_master['redis_port'], 0, 8191, tps)
p = util.exec_proc_async(util.cluster_util_dir(src_master['id']), cmd,
True, None, subprocess.PIPE, None)
ret = p.wait()
for line in p.stdout:
if line.find("Checkpoint Sequence Number:") != -1:
util.log("seqnumber : " + line[line.rfind(":") + 1:])
seq = int(line[line.rfind(":") + 1:])
util.log(">>>" + str(line.rstrip()))
self.assertEqual(0, ret)
util.log(">>> end remote checkpoint and load (%s)" % time.asctime())
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis,
'beforeCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis,
'beforeCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis,
'S3:beforeCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis,
'S3:beforeCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
# bgsave for testing later about recovery during migration
util.log(
">>> bgsave for testing later about recovery during migration (%s)"
% time.asctime())
cmd = 'bgsave\r\n'
dst_redis.write(cmd)
res = dst_redis.read_until('\r\n')
self.assertEquals(res, '+Background saving started\r\n')
ts = time.time()
self.setExpireKey(src_redis, 'afterCheckpoint~afterCheckpoint:expired',
10)
self.setExpireKey(src_redis, 'afterCheckpoint~afterCheckpoint:persist',
20)
self.setExpireS3Key(src_redis,
'S3:afterCheckpoint~afterCheckpoint:expired', 10)
self.setExpireS3Key(src_redis,
'S3:afterCheckpoint~afterCheckpoint:persist', 20)
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis,
'afterCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis,
'afterCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis,
'S3:afterCheckpoint~afterCheckpoint:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis,
'S3:afterCheckpoint~afterCheckpoint:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(src_redis, 'afterCheckpoint~duringCatchup:expired',
10)
self.setExpireKey(src_redis, 'afterCheckpoint~duringCatchup:persist',
100)
self.setExpireS3Key(src_redis,
'S3:afterCheckpoint~duringCatchup:expired', 10)
self.setExpireS3Key(src_redis,
'S3:afterCheckpoint~duringCatchup:persist', 100)
# remote catchup (smr log migration)
util.log(">>> start remote catchup (%s)" % time.asctime())
dst_host = dst_master['ip']
dst_smr_port = dst_master['smr_base_port']
rle = '1 8192'
num_part = 8192
smr.write('migrate start %s %d %d %d %d %s\r\n' %
(dst_host, dst_smr_port, seq, tps, num_part, rle))
response = smr.read_until('\r\n')
if response[:3] != '+OK':
util.log('failed to execute migrate start command, response:%s' %
response)
return False
while True:
smr.write('migrate info\r\n')
response = smr.read_until('\r\n')
seqs = response.split()
logseq = int(seqs[1].split(':')[1])
mig = int(seqs[2].split(':')[1])
util.log('migrate info: %s' % response)
if (logseq - mig < 500000):
util.log('Remote catchup almost done. try mig2pc')
break
time.sleep(1)
util.log(">>> sleep until 90 sec pass")
self.assertFalse(time.time() - ts >= 90)
time.sleep(90 - (time.time() - ts))
res = self.persistKey(src_redis,
'afterCheckpoint~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis,
'afterCheckpoint~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis,
'S3:afterCheckpoint~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis,
'S3:afterCheckpoint~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(src_redis, 'duringCatchup~duringCatchup:expired', 10)
self.setExpireKey(src_redis, 'duringCatchup~duringCatchup:persist', 20)
self.setExpireS3Key(src_redis,
'S3:duringCatchup~duringCatchup:expired', 10)
self.setExpireS3Key(src_redis,
'S3:duringCatchup~duringCatchup:persist', 20)
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(src_redis, 'duringCatchup~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(src_redis, 'duringCatchup~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(src_redis,
'S3:duringCatchup~duringCatchup:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(src_redis,
'S3:duringCatchup~duringCatchup:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(src_redis, 'duringCatchup~afterMig2pc:expired', 10)
self.setExpireKey(src_redis, 'duringCatchup~afterMig2pc:persist', 20)
self.setExpireS3Key(src_redis, 'S3:duringCatchup~afterMig2pc:expired',
10)
self.setExpireS3Key(src_redis, 'S3:duringCatchup~afterMig2pc:persist',
20)
util.log(">>> remote catchup phase almost done (%s)" % time.asctime())
# mig2pc
util.log(">>> start mig2pc (%s)" % time.asctime())
cmd = 'mig2pc %s %d %d %d %d' % (self.cluster['cluster_name'],
src_pg_id, dst_pg_id, 0, 8191)
result = util.cm_command(leader_cm['ip'], leader_cm['cm_port'], cmd)
util.log('mig2pc result : ' + result)
if not result.startswith('{"state":"success","msg":"+OK"}\r\n'):
util.log('failed to execute mig2pc command, result:%s' % result)
return False
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(dst_redis, 'duringCatchup~afterMig2pc:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(dst_redis, 'duringCatchup~afterMig2pc:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(dst_redis,
'S3:duringCatchup~afterMig2pc:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(dst_redis,
'S3:duringCatchup~afterMig2pc:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
self.setExpireKey(dst_redis, 'afterMig2pc~migrateEnd:expired', 10)
self.setExpireKey(dst_redis, 'afterMig2pc~migrateEnd:persist', 20)
self.setExpireS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:expired', 10)
self.setExpireS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:persist', 20)
# finish migration
smr.write('migrate interrupt\r\n')
response = smr.read_until('\r\n')
util.log('migrate interrupt: %s' % response)
smr.disconnect()
# notify dst_redis of migration end
util.log(">>> notify dst_redis of migration end (%s)" % time.asctime())
cmd = 'migconf migend\r\n'
dst_redis.write(cmd)
res = dst_redis.read_until('\r\n')
self.assertEquals(res, '+OK\r\n')
cmd = 'migconf clearstart %d-%d\r\n' % (0, 8191)
src_redis.write(cmd)
res = src_redis.read_until('\r\n')
self.assertEquals(res, '+OK\r\n')
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
res = self.persistKey(dst_redis, 'afterMig2pc~migrateEnd:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistKey(dst_redis, 'afterMig2pc~migrateEnd:expired')
self.assertEquals(res, ":0\r\n")
res = self.persistS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:persist')
self.assertEquals(res, ":1\r\n")
res = self.persistS3Key(dst_redis, 'S3:afterMig2pc~migrateEnd:expired')
self.assertEquals(res, ":0\r\n")
ts = time.time()
util.log(">>> sleep until 15 sec pass")
self.assertFalse(time.time() - ts >= 15)
time.sleep(15 - (time.time() - ts))
self.assertTrue(
self.isExist(dst_redis,
'beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis,
'beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis,
'beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis,
'beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:persist'))
self.assertFalse(
self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~duringCatchup:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~duringCatchup:expired'))
self.assertTrue(
self.isExist(dst_redis, 'duringCatchup~duringCatchup:persist'))
self.assertFalse(
self.isExist(dst_redis, 'duringCatchup~duringCatchup:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:duringCatchup~duringCatchup:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:duringCatchup~duringCatchup:expired'))
self.assertTrue(
self.isExist(dst_redis, 'duringCatchup~afterMig2pc:persist'))
self.assertFalse(
self.isExist(dst_redis, 'duringCatchup~afterMig2pc:expired'))
self.assertTrue(
self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:persist'))
self.assertFalse(
self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:expired'))
self.assertTrue(
self.isExist(dst_redis, 'afterMig2pc~migrateEnd:persist'))
self.assertFalse(
self.isExist(dst_redis, 'afterMig2pc~migrateEnd:expired'))
self.assertTrue(
self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:persist'))
self.assertFalse(
self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:expired'))
# remote partial checkpoint
util.log(">>> start rangedel (%s)" % time.asctime())
cmd = "./cluster-util --rangedel %s %d %d-%d %d" % (
src_master['ip'], src_master['redis_port'], 0, 8191, tps)
p = util.exec_proc_async(util.cluster_util_dir(src_master['id']), cmd,
True, None, subprocess.PIPE, None)
ret = p.wait()
for line in p.stdout:
util.log(">>>" + str(line.rstrip()))
cmd = 'migconf clearend\r\n'
src_redis.write(cmd)
res = src_redis.read_until('\r\n')
self.assertEqual(res, '+OK\r\n')
time.sleep(5) # generate load for 5 sec
# check consistency of load_generator
for i in range(len(load_gen_thrd_list)):
load_gen_thrd_list[i].quit()
for i in range(len(load_gen_thrd_list)):
load_gen_thrd_list[i].join()
self.assertTrue(load_gen_thrd_list[i].isConsistent(),
'Inconsistent after migration')
# kill dst_redis and recover from bgsave
util.log(">>> kill dst_redis and recover from bgsave (%s)" %
time.asctime())
dst_redis.disconnect()
ret = testbase.request_to_shutdown_redis(dst_master)
self.assertEquals(ret, 0, 'failed to shutdown redis')
ret = testbase.request_to_shutdown_smr(dst_master)
self.assertEquals(ret, 0, 'failed to shutdown smr')
time.sleep(5)
testbase.request_to_start_smr(dst_master)
self.assertEqual(ret, 0,
'failed to start smr, server:%d' % dst_master['id'])
ret = testbase.request_to_start_redis(dst_master)
self.assertEqual(ret, 0,
'failed to start redis, server:%d' % dst_master['id'])
ret = testbase.wait_until_finished_to_set_up_role(dst_master)
self.assertEquals(
ret, 0, 'failed to role change. server:%d' % (dst_master['id']))
dst_redis = redis_mgmt.Redis(dst_master['id'])
ret = dst_redis.connect(dst_master['ip'], dst_master['redis_port'])
self.assertEquals(ret, 0, 'failed to connect to redis')
self.assertTrue(
self.isExist(dst_redis,
'beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis,
'beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis,
'beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis,
'beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis, 'afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:persist'))
self.assertFalse(
self.isExist(dst_redis, 'afterCheckpoint~duringCatchup:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~duringCatchup:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:afterCheckpoint~duringCatchup:expired'))
self.assertTrue(
self.isExist(dst_redis, 'duringCatchup~duringCatchup:persist'))
self.assertFalse(
self.isExist(dst_redis, 'duringCatchup~duringCatchup:expired'))
self.assertTrue(
self.isS3Exist(dst_redis,
'S3:duringCatchup~duringCatchup:persist'))
self.assertFalse(
self.isS3Exist(dst_redis,
'S3:duringCatchup~duringCatchup:expired'))
self.assertTrue(
self.isExist(dst_redis, 'duringCatchup~afterMig2pc:persist'))
self.assertFalse(
self.isExist(dst_redis, 'duringCatchup~afterMig2pc:expired'))
self.assertTrue(
self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:persist'))
self.assertFalse(
self.isS3Exist(dst_redis, 'S3:duringCatchup~afterMig2pc:expired'))
self.assertTrue(
self.isExist(dst_redis, 'afterMig2pc~migrateEnd:persist'))
self.assertFalse(
self.isExist(dst_redis, 'afterMig2pc~migrateEnd:expired'))
self.assertTrue(
self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:persist'))
self.assertFalse(
self.isS3Exist(dst_redis, 'S3:afterMig2pc~migrateEnd:expired'))
self.getS3TTL(dst_redis, 'S3:PermanentKey')
# kill dst_slave redis and recover without dump file
util.log(">>> kill dst_redis and recover without dump file (%s)" %
time.asctime())
dst_slave = util.get_server_by_role_and_pg(self.cluster['servers'],
'slave', dst_pg_id)
ret = testbase.request_to_shutdown_redis(dst_slave)
self.assertEquals(ret, 0, 'failed to shutdown redis')
ret = testbase.request_to_shutdown_smr(dst_slave)
self.assertEquals(ret, 0, 'failed to shutdown smr')
time.sleep(5)
testbase.request_to_start_smr(dst_slave)
self.assertEqual(ret, 0,
'failed to start smr, server:%d' % dst_slave['id'])
ret = testbase.request_to_start_redis(dst_slave)
self.assertEqual(ret, 0,
'failed to start redis, server:%d' % dst_slave['id'])
ret = testbase.wait_until_finished_to_set_up_role(dst_slave)
self.assertEquals(
ret, 0, 'failed to role change. server:%d' % (dst_slave['id']))
dst_redis_slave = redis_mgmt.Redis(dst_slave['id'])
ret = dst_redis_slave.connect(dst_slave['ip'], dst_slave['redis_port'])
self.assertEquals(ret, 0, 'failed to connect to redis')
self.assertTrue(
self.isExist(dst_redis_slave,
'beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis_slave,
'beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:beforeCheckpoint~beforeCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:beforeCheckpoint~beforeCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis_slave,
'beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis_slave,
'beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:beforeCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:beforeCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis_slave,
'afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isExist(dst_redis_slave,
'afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:afterCheckpoint~afterCheckpoint:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:afterCheckpoint~afterCheckpoint:expired'))
self.assertTrue(
self.isExist(dst_redis_slave,
'afterCheckpoint~duringCatchup:persist'))
self.assertFalse(
self.isExist(dst_redis_slave,
'afterCheckpoint~duringCatchup:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:afterCheckpoint~duringCatchup:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:afterCheckpoint~duringCatchup:expired'))
self.assertTrue(
self.isExist(dst_redis_slave,
'duringCatchup~duringCatchup:persist'))
self.assertFalse(
self.isExist(dst_redis_slave,
'duringCatchup~duringCatchup:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:duringCatchup~duringCatchup:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:duringCatchup~duringCatchup:expired'))
self.assertTrue(
self.isExist(dst_redis_slave, 'duringCatchup~afterMig2pc:persist'))
self.assertFalse(
self.isExist(dst_redis_slave, 'duringCatchup~afterMig2pc:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:duringCatchup~afterMig2pc:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:duringCatchup~afterMig2pc:expired'))
self.assertTrue(
self.isExist(dst_redis_slave, 'afterMig2pc~migrateEnd:persist'))
self.assertFalse(
self.isExist(dst_redis_slave, 'afterMig2pc~migrateEnd:expired'))
self.assertTrue(
self.isS3Exist(dst_redis_slave,
'S3:afterMig2pc~migrateEnd:persist'))
self.assertFalse(
self.isS3Exist(dst_redis_slave,
'S3:afterMig2pc~migrateEnd:expired'))
self.getS3TTL(dst_redis_slave, 'S3:PermanentKey')
# Go back to initial configuration
self.assertTrue(
util.migration(self.cluster, dst_pg_id, src_pg_id, 0, 8191, 40000),
'failed to rollback migration')
def numOpenFds(self, pid):
p = util.exec_proc_async(util.cluster_util_dir(0),
'ls /proc/%s/fd | wc -l' % pid, True, None,
subprocess.PIPE, None)
ret = p.wait()
return p.stdout.read()[:-1]