1
0
mirror of https://github.com/postgrespro/pg_probackup.git synced 2024-12-11 11:41:33 +02:00
pg_probackup/tests/page.py

526 lines
18 KiB
Python
Raw Normal View History

2017-07-12 16:28:28 +02:00
import os
import unittest
from .helpers.ptrack_helpers import ProbackupTest, ProbackupException
from datetime import datetime, timedelta
2018-01-16 17:46:55 +02:00
import subprocess
import time
2017-07-12 16:28:28 +02:00
module_name = 'page'
class PageBackupTest(ProbackupTest, unittest.TestCase):
2018-01-16 17:46:55 +02:00
# @unittest.skip("skip")
def test_page_vacuum_truncate(self):
"""
make node, create table, take full backup,
delete last 3 pages, vacuum relation,
take page backup, take second page backup,
restore last page backup and check data correctness
"""
2018-01-16 17:46:55 +02:00
fname = self.id().split('.')[3]
backup_dir = os.path.join(self.tmp_path, module_name, fname, 'backup')
node = self.make_simple_node(
base_dir="{0}/{1}/node".format(module_name, fname),
set_replication=True,
initdb_params=['--data-checksums'],
pg_options={
'wal_level': 'replica',
'max_wal_senders': '2',
'checkpoint_timeout': '300s',
'autovacuum': 'off'
2018-01-16 17:46:55 +02:00
}
)
node_restored = self.make_simple_node(
base_dir="{0}/{1}/node_restored".format(module_name, fname),
)
self.init_pb(backup_dir)
self.add_instance(backup_dir, 'node', node)
self.set_archiving(backup_dir, 'node', node)
node_restored.cleanup()
node.start()
self.create_tblspace_in_node(node, 'somedata')
node.safe_psql(
"postgres",
"create sequence t_seq; "
"create table t_heap tablespace somedata as select i as id, "
"md5(i::text) as text, "
"md5(repeat(i::text,10))::tsvector as tsvector "
"from generate_series(0,1024) i;"
)
node.safe_psql(
"postgres",
"vacuum t_heap"
)
self.backup_node(backup_dir, 'node', node)
node.safe_psql(
"postgres",
"delete from t_heap where ctid >= '(11,0)'"
)
node.safe_psql(
"postgres",
"vacuum t_heap"
)
self.backup_node(
backup_dir, 'node', node, backup_type='page',
options=['--log-level-file=verbose']
)
self.backup_node(
backup_dir, 'node', node, backup_type='page'
2018-01-16 17:46:55 +02:00
)
if self.paranoia:
pgdata = self.pgdata_content(node.data_dir)
old_tablespace = self.get_tblspace_path(node, 'somedata')
new_tablespace = self.get_tblspace_path(node_restored, 'somedata_new')
self.restore_node(
backup_dir, 'node', node_restored,
options=[
"-j", "4",
"-T", "{0}={1}".format(old_tablespace, new_tablespace),
"--recovery-target-action=promote"]
2018-01-16 17:46:55 +02:00
)
# Physical comparison
if self.paranoia:
pgdata_restored = self.pgdata_content(node_restored.data_dir)
self.compare_pgdata(pgdata, pgdata_restored)
node_restored.append_conf(
"postgresql.auto.conf", "port = {0}".format(node_restored.port))
node_restored.start()
while node_restored.safe_psql(
"postgres", "select pg_is_in_recovery()") == 't\n':
time.sleep(1)
# Logical comparison
result1 = node.safe_psql(
"postgres",
"select * from t_heap"
)
result2 = node_restored.safe_psql(
"postgres",
"select * from t_heap"
)
self.assertEqual(result1, result2)
2018-01-16 17:46:55 +02:00
# Clean after yourself
self.del_test_dir(module_name, fname)
2017-07-12 16:28:28 +02:00
# @unittest.skip("skip")
def test_page_stream(self):
"""
make archive node, take full and page stream backups,
restore them and check data correctness
"""
2017-07-12 16:28:28 +02:00
self.maxDiff = None
fname = self.id().split('.')[3]
backup_dir = os.path.join(self.tmp_path, module_name, fname, 'backup')
2018-01-16 17:46:55 +02:00
node = self.make_simple_node(
base_dir="{0}/{1}/node".format(module_name, fname),
2017-07-12 16:28:28 +02:00
set_replication=True,
initdb_params=['--data-checksums'],
2018-01-16 17:46:55 +02:00
pg_options={
'wal_level': 'replica',
'max_wal_senders': '2',
'checkpoint_timeout': '30s',
'ptrack_enable': 'on'}
2017-07-12 16:28:28 +02:00
)
self.init_pb(backup_dir)
self.add_instance(backup_dir, 'node', node)
self.set_archiving(backup_dir, 'node', node)
node.start()
# FULL BACKUP
node.safe_psql(
"postgres",
2018-01-16 17:46:55 +02:00
"create table t_heap as select i as id, md5(i::text) as text, "
"md5(i::text)::tsvector as tsvector "
"from generate_series(0,100) i")
2017-07-12 16:28:28 +02:00
full_result = node.execute("postgres", "SELECT * FROM t_heap")
2018-01-16 17:46:55 +02:00
full_backup_id = self.backup_node(
backup_dir, 'node', node,
backup_type='full', options=['--stream'])
2017-07-12 16:28:28 +02:00
2018-01-16 17:46:55 +02:00
# PAGE BACKUP
2017-07-12 16:28:28 +02:00
node.safe_psql(
"postgres",
2018-01-16 17:46:55 +02:00
"insert into t_heap select i as id, md5(i::text) as text, "
"md5(i::text)::tsvector as tsvector "
"from generate_series(100,200) i")
2017-07-12 16:28:28 +02:00
page_result = node.execute("postgres", "SELECT * FROM t_heap")
2018-01-16 17:46:55 +02:00
page_backup_id = self.backup_node(
backup_dir, 'node', node,
backup_type='page', options=['--stream'])
2017-07-12 16:28:28 +02:00
# Drop Node
node.cleanup()
# Check full backup
2018-01-16 17:46:55 +02:00
self.assertIn(
"INFO: Restore of backup {0} completed.".format(full_backup_id),
self.restore_node(
backup_dir, 'node', node,
backup_id=full_backup_id, options=["-j", "4"]),
'\n Unexpected Error Message: {0}\n'
' CMD: {1}'.format(repr(self.output), self.cmd))
2017-07-12 16:28:28 +02:00
node.start()
full_result_new = node.execute("postgres", "SELECT * FROM t_heap")
self.assertEqual(full_result, full_result_new)
node.cleanup()
# Check page backup
2018-01-16 17:46:55 +02:00
self.assertIn(
"INFO: Restore of backup {0} completed.".format(page_backup_id),
self.restore_node(
backup_dir, 'node', node,
backup_id=page_backup_id, options=["-j", "4"]),
'\n Unexpected Error Message: {0}\n'
' CMD: {1}'.format(repr(self.output), self.cmd))
2017-07-12 16:28:28 +02:00
node.start()
page_result_new = node.execute("postgres", "SELECT * FROM t_heap")
self.assertEqual(page_result, page_result_new)
node.cleanup()
# Clean after yourself
self.del_test_dir(module_name, fname)
# @unittest.skip("skip")
def test_page_archive(self):
"""
make archive node, take full and page archive backups,
restore them and check data correctness
"""
2017-07-12 16:28:28 +02:00
self.maxDiff = None
fname = self.id().split('.')[3]
backup_dir = os.path.join(self.tmp_path, module_name, fname, 'backup')
node = self.make_simple_node(
base_dir="{0}/{1}/node".format(module_name, fname),
2017-07-12 16:28:28 +02:00
set_replication=True,
initdb_params=['--data-checksums'],
pg_options={
'wal_level': 'replica',
'max_wal_senders': '2',
'checkpoint_timeout': '30s',
'ptrack_enable': 'on'}
2017-07-12 16:28:28 +02:00
)
self.init_pb(backup_dir)
self.add_instance(backup_dir, 'node', node)
self.set_archiving(backup_dir, 'node', node)
node.start()
# FULL BACKUP
node.safe_psql(
"postgres",
"create table t_heap as select i as id, md5(i::text) as text, "
"md5(i::text)::tsvector as tsvector from generate_series(0,1) i")
2017-07-12 16:28:28 +02:00
full_result = node.execute("postgres", "SELECT * FROM t_heap")
full_backup_id = self.backup_node(
backup_dir, 'node', node, backup_type='full')
2017-07-12 16:28:28 +02:00
# PAGE BACKUP
2017-07-12 16:28:28 +02:00
node.safe_psql(
"postgres",
"insert into t_heap select i as id, "
"md5(i::text) as text, md5(i::text)::tsvector as tsvector "
"from generate_series(0,2) i")
2017-07-12 16:28:28 +02:00
page_result = node.execute("postgres", "SELECT * FROM t_heap")
page_backup_id = self.backup_node(
backup_dir, 'node', node, backup_type='page')
2017-07-12 16:28:28 +02:00
# Drop Node
node.cleanup()
# Restore and check full backup
self.assertIn("INFO: Restore of backup {0} completed.".format(
full_backup_id),
self.restore_node(
backup_dir, 'node', node,
backup_id=full_backup_id,
options=["-j", "4"]),
'\n Unexpected Error Message: {0}\n CMD: {1}'.format(
repr(self.output), self.cmd))
2017-07-12 16:28:28 +02:00
node.start()
full_result_new = node.execute("postgres", "SELECT * FROM t_heap")
self.assertEqual(full_result, full_result_new)
node.cleanup()
# Restore and check page backup
self.assertIn(
"INFO: Restore of backup {0} completed.".format(page_backup_id),
self.restore_node(
backup_dir, 'node', node,
backup_id=page_backup_id,
options=["-j", "4"]),
'\n Unexpected Error Message: {0}\n CMD: {1}'.format(
repr(self.output), self.cmd))
2017-07-12 16:28:28 +02:00
node.start()
page_result_new = node.execute("postgres", "SELECT * FROM t_heap")
self.assertEqual(page_result, page_result_new)
node.cleanup()
# Clean after yourself
self.del_test_dir(module_name, fname)
2017-11-09 11:45:04 +02:00
# @unittest.skip("skip")
def test_page_multiple_segments(self):
"""
Make node, create table with multiple segments,
write some data to it, check page and data correctness
"""
2017-11-09 11:45:04 +02:00
fname = self.id().split('.')[3]
backup_dir = os.path.join(self.tmp_path, module_name, fname, 'backup')
node = self.make_simple_node(
base_dir="{0}/{1}/node".format(module_name, fname),
2017-11-09 11:45:04 +02:00
set_replication=True,
initdb_params=['--data-checksums'],
pg_options={
'wal_level': 'replica',
'max_wal_senders': '2',
'ptrack_enable': 'on',
'fsync': 'off',
'shared_buffers': '1GB',
'maintenance_work_mem': '1GB',
'autovacuum': 'off',
'full_page_writes': 'off'
}
2017-11-09 11:45:04 +02:00
)
self.init_pb(backup_dir)
self.add_instance(backup_dir, 'node', node)
self.set_archiving(backup_dir, 'node', node)
node.start()
self.create_tblspace_in_node(node, 'somedata')
# CREATE TABLE
node.pgbench_init(scale=100, options=['--tablespace=somedata'])
# FULL BACKUP
self.backup_node(backup_dir, 'node', node)
# PGBENCH STUFF
pgbench = node.pgbench(options=['-T', '50', '-c', '1', '--no-vacuum'])
2017-11-09 11:45:04 +02:00
pgbench.wait()
node.safe_psql("postgres", "checkpoint")
# GET LOGICAL CONTENT FROM NODE
result = node.safe_psql("postgres", "select * from pgbench_accounts")
# PAGE BACKUP
self.backup_node(
backup_dir, 'node', node, backup_type='page',
options=["--log-level-file=verbose"])
2017-11-09 11:45:04 +02:00
# GET PHYSICAL CONTENT FROM NODE
pgdata = self.pgdata_content(node.data_dir)
# RESTORE NODE
restored_node = self.make_simple_node(
base_dir="{0}/{1}/restored_node".format(module_name, fname))
2017-11-09 11:45:04 +02:00
restored_node.cleanup()
tblspc_path = self.get_tblspace_path(node, 'somedata')
tblspc_path_new = self.get_tblspace_path(
restored_node, 'somedata_restored')
2017-11-09 11:45:04 +02:00
self.restore_node(
backup_dir, 'node', restored_node,
options=[
"-j", "4",
"--recovery-target-action=promote",
"-T", "{0}={1}".format(tblspc_path, tblspc_path_new)])
2017-11-09 11:45:04 +02:00
# GET PHYSICAL CONTENT FROM NODE_RESTORED
pgdata_restored = self.pgdata_content(restored_node.data_dir)
# START RESTORED NODE
restored_node.append_conf(
"postgresql.auto.conf", "port = {0}".format(restored_node.port))
2017-11-09 11:45:04 +02:00
restored_node.start()
while restored_node.safe_psql(
"postgres", "select pg_is_in_recovery()") == 't\n':
2017-11-09 11:45:04 +02:00
time.sleep(1)
result_new = restored_node.safe_psql(
"postgres", "select * from pgbench_accounts")
2017-11-09 11:45:04 +02:00
# COMPARE RESTORED FILES
self.assertEqual(result, result_new, 'data is lost')
if self.paranoia:
self.compare_pgdata(pgdata, pgdata_restored)
# Clean after yourself
self.del_test_dir(module_name, fname)
# @unittest.skip("skip")
def test_page_delete(self):
"""
Make node, create tablespace with table, take full backup,
delete everything from table, vacuum table, take page backup,
restore page backup, compare .
"""
fname = self.id().split('.')[3]
backup_dir = os.path.join(self.tmp_path, module_name, fname, 'backup')
node = self.make_simple_node(
base_dir="{0}/{1}/node".format(module_name, fname),
set_replication=True, initdb_params=['--data-checksums'],
pg_options={
'wal_level': 'replica',
'max_wal_senders': '2',
'checkpoint_timeout': '30s',
'autovacuum': 'off'
}
)
self.init_pb(backup_dir)
self.add_instance(backup_dir, 'node', node)
self.set_archiving(backup_dir, 'node', node)
node.start()
self.create_tblspace_in_node(node, 'somedata')
# FULL backup
self.backup_node(backup_dir, 'node', node)
node.safe_psql(
"postgres",
"create table t_heap tablespace somedata as select i as id,"
" md5(i::text) as text, md5(i::text)::tsvector as tsvector"
" from generate_series(0,100) i"
)
node.safe_psql(
"postgres",
"delete from t_heap"
)
node.safe_psql(
"postgres",
"vacuum t_heap"
)
# PAGE BACKUP
self.backup_node(
backup_dir, 'node', node, backup_type='page')
if self.paranoia:
pgdata = self.pgdata_content(node.data_dir)
# RESTORE
node_restored = self.make_simple_node(
base_dir="{0}/{1}/node_restored".format(module_name, fname)
)
node_restored.cleanup()
self.restore_node(
backup_dir, 'node', node_restored,
options=[
"-j", "4",
"-T", "{0}={1}".format(
self.get_tblspace_path(node, 'somedata'),
self.get_tblspace_path(node_restored, 'somedata'))
]
)
# GET RESTORED PGDATA AND COMPARE
if self.paranoia:
pgdata_restored = self.pgdata_content(node_restored.data_dir)
self.compare_pgdata(pgdata, pgdata_restored)
# START RESTORED NODE
node_restored.append_conf(
'postgresql.auto.conf', 'port = {0}'.format(node_restored.port))
node_restored.start()
# Clean after yourself
self.del_test_dir(module_name, fname)
# @unittest.skip("skip")
def test_page_delete_1(self):
"""
Make node, create tablespace with table, take full backup,
delete everything from table, vacuum table, take page backup,
restore page backup, compare .
"""
fname = self.id().split('.')[3]
backup_dir = os.path.join(self.tmp_path, module_name, fname, 'backup')
node = self.make_simple_node(
base_dir="{0}/{1}/node".format(module_name, fname),
set_replication=True, initdb_params=['--data-checksums'],
pg_options={
'wal_level': 'replica',
'max_wal_senders': '2',
'checkpoint_timeout': '30s',
'autovacuum': 'off'
}
)
self.init_pb(backup_dir)
self.add_instance(backup_dir, 'node', node)
self.set_archiving(backup_dir, 'node', node)
node.start()
self.create_tblspace_in_node(node, 'somedata')
node.safe_psql(
"postgres",
"create table t_heap tablespace somedata as select i as id,"
" md5(i::text) as text, md5(i::text)::tsvector as tsvector"
" from generate_series(0,100) i"
)
# FULL backup
self.backup_node(backup_dir, 'node', node)
node.safe_psql(
"postgres",
"delete from t_heap"
)
node.safe_psql(
"postgres",
"vacuum t_heap"
)
# PAGE BACKUP
self.backup_node(
backup_dir, 'node', node, backup_type='page')
if self.paranoia:
pgdata = self.pgdata_content(node.data_dir)
# RESTORE
node_restored = self.make_simple_node(
base_dir="{0}/{1}/node_restored".format(module_name, fname)
)
node_restored.cleanup()
self.restore_node(
backup_dir, 'node', node_restored,
options=[
"-j", "4",
"-T", "{0}={1}".format(
self.get_tblspace_path(node, 'somedata'),
self.get_tblspace_path(node_restored, 'somedata'))
]
)
# GET RESTORED PGDATA AND COMPARE
if self.paranoia:
pgdata_restored = self.pgdata_content(node_restored.data_dir)
self.compare_pgdata(pgdata, pgdata_restored)
# START RESTORED NODE
node_restored.append_conf(
'postgresql.auto.conf', 'port = {0}'.format(node_restored.port))
node_restored.start()
# Clean after yourself
self.del_test_dir(module_name, fname)