nobase_noinst_SCRIPTS += $(wildcard python/tests-infra/*.py)
nobase_noinst_SCRIPTS += $(wildcard python/tests/suite_dlc/*.py)
nobase_noinst_SCRIPTS += $(wildcard python/tests/suite_samples/*.py)
+nobase_noinst_SCRIPTS += $(wildcard python/tests/suite_dynamic-nids/*.py)
EXTRA_DIST=$(nobase_noinst_DATA) $(nobase_noinst_SCRIPTS)
noinst_testdir = $(libdir)/lustre/tests/lutf
--- /dev/null
+from lnet_cleanup import clean_lnet
+from lustre_cleanup import clean_lustre
+
+def lutf_clean_setup():
+ clean_lustre()
+ clean_lnet()
--- /dev/null
+"""
+@PRIMARY: LU-10360-01
+@PRIMARY_DESC: Eanble/Disable dynamic NIDs via dynamic_nids parameter
+@SECONDARY: LU-10360-02
+@DESIGN: Standard Lustre sysfs module parameters setting
+@TESTCASE:
+On the client and on the MDS
+ lctl set_param mgc.*.dynamic_nids=1
+ lctl get_param mgc.*.dynamic_nids
+ Check the value is 1
+On the client and on the MDS
+ lctl set_param mgc.*.dynamic_nids=0
+ lctl get_param mgc.*.dynamic_nids
+ Check the value is 0
+"""
+
+import os, re, yaml, logging
+from time import sleep
+import lnetconfig
+from lutf import agents, me
+from lutf_basetest import *
+from lnet import TheLNet
+from lutf_exception import *
+from lnet_selftest import LNetSelfTest
+from lnet_helpers import LNetHelpers
+from lustre_node import SimpleLustreNode
+from lustre_roles import *
+from lustre_fs import SimpleLustreFS
+import lutf_common_def as common
+
+def set_check_dynamic_nids(node, prev, new):
+ dn = node.get_dynamic_nids()
+ if dn != prev:
+ return False, prev
+ node.set_dynamic_nids(new)
+ dn = node.get_dynamic_nids()
+ if dn != new:
+ return False, new
+ return True, new
+
+def run():
+ la = agents.keys()
+ if len(la) < 3:
+ return lutfrc(LUTF_TEST_SKIP,
+ msg="Not enough agents to setup a lustre FS. 3 needed %d found" % len(la))
+ try:
+ lustrefs = SimpleLustreFS(la)
+ MGSs = lustrefs.get_mgs_nodes()
+ OSSs = lustrefs.get_oss_nodes()
+ clients = lustrefs.get_client_nodes()
+ if len(MGSs) < 1 or len(OSSs) < 1 or len(clients) < 1:
+ return lutfrc(LUTF_TEST_SKIP, "Unexpected cluster: MGSs %d, OSSs %d, clients %d" % (len(MGSs), len(OSSs), len(clients)))
+ lustrefs.configure_nets()
+ lustrefs.mount_servers(num_mgs=1, num_oss=1)
+ lustrefs.mount_clients(num=1)
+ rc, value = set_check_dynamic_nids(MGSs[0], 0, 1)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, value = set_check_dynamic_nids(MGSs[0], 1, 0)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, value = set_check_dynamic_nids(clients[0], 0, 1)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, value = set_check_dynamic_nids(clients[0], 1, 0)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ return lutfrc(LUTF_TEST_PASS)
+ except Exception as e:
+ raise e
+
--- /dev/null
+"""
+@PRIMARY: LU-10360-03
+@PRIMARY_DESC: Allow clients to continue using servers which have
+changed their IP address during a boot cycle
+@DESIGN:
+An IR log is sent with the current NID information of the server. When the
+client receives the IR log it checks the entry there against what it has already
+stored from the llog.
+ If the entry is not there, then add a new connection to the import
+ If the entry is there but the NID list is different, then update the NID
+ information with the latest NID information provided in the IR log.
+Since allowing new servers NIDs previously unknown during the initial mount to
+be used, it could be considered a security risk on some sites.
+ Add a new File system level module parameter to enable this feature.
+ The feature is disabled by default. To Enable the feature:
+ lctl set_param mgc.*.dynamic_nids=1
+@TESTCASE:
+ Setup a file system consistent of:
+ 1 node: MGS/MDS
+ 1 node: OSS should have 2 IP addresses
+ 1 node: Client
+ Mount the File System and the client
+ Use only one of the IP addresses on the OSS
+ Enable the dynamic_nids feature on both the MDS and the client
+ Write multiple files to the FS
+ Read back the files and make sure they are correct
+ unmount the OSS and remove lustre modules
+ Change the OSS configuration to use the other IP address for NID
+ load lustre on the OSS and re mount
+ LS the file system.
+ We should get the exact same list of files
+ Write more files and read them back.
+ the write/read operation should succeed
+"""
+
+import os, re, yaml, logging
+from time import sleep
+import lnetconfig
+from lutf import agents, me
+from lutf_basetest import *
+from lnet import TheLNet
+from lutf_exception import *
+from lnet_selftest import LNetSelfTest
+from lnet_helpers import LNetHelpers
+from lustre_node import SimpleLustreNode
+from lustre_roles import *
+from lustre_fs import SimpleLustreFS
+import lutf_common_def as common
+from lutf_utils import *
+
+def set_check_dynamic_nids(node, prev, new):
+ dn = node.get_dynamic_nids()
+ if dn != prev:
+ return False, prev
+ node.set_dynamic_nids(new)
+ dn = node.get_dynamic_nids()
+ if dn != new:
+ return False, new
+ return True, new
+
+def write_verify_to_fs(lustrefs):
+ data = generate_random_bytes(1048576)
+ files = []
+ # write 3 files
+ for i in range(0, 3):
+ files.append(lustrefs.write(0, data, thread=True))
+ # read and verify each of the files written
+ for f in files:
+ if not lustrefs.read_verify(0, data, f, thread=True):
+ return False, f
+ return True, None
+
+def run():
+ la = agents.keys()
+ if len(la) < 3:
+ return lutfrc(LUTF_TEST_SKIP,
+ msg="Not enough agents to setup a lustre FS. 3 needed %d found" % len(la))
+ try:
+ lustrefs = SimpleLustreFS(la)
+ # configure the OSS to use the first IP address
+ MGSs = lustrefs.get_mgs_nodes()
+ OSSs = lustrefs.get_oss_nodes()
+ clients = lustrefs.get_client_nodes()
+ if len(MGSs) < 1 or len(OSSs) < 1 or len(clients) < 1:
+ return lutfrc(LUTF_TEST_SKIP, "Unexpected cluster: MGSs %d, OSSs %d, clients %d" % (len(MGSs), len(OSSs), len(clients)))
+
+ oss = lustrefs.get_oss_nodes()
+ oss_intfs = oss[0].list_intfs()
+ if len(oss_intfs) <= 1:
+ return lutfrc(LUTF_TEST_SKIP, "not enough interfaces on oss %s (%s)" % (oss[0].get_node_name(), oss[0].get_node_hostname()))
+ oss_net = {'tcp': [oss_intfs[0]]}
+ lustrefs.configure_nets(oss=oss_net)
+ lustrefs.mount_servers(num_mgs=1, num_oss=1)
+ lustrefs.mount_clients(num=1)
+ rc, value = set_check_dynamic_nids(MGSs[0], 0, 1)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, value = set_check_dynamic_nids(clients[0], 0, 1)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, f = write_verify_to_fs(lustrefs)
+ if not rc:
+ lutfrc(LUTF_TEST_FAIL, "Failed to verify data on " + f)
+ logging.debug("unconfigure lustre on: " + oss[0].get_node_name())
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[1]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ sleep(10)
+ logging.debug("configure lustre on: " + oss[0].get_node_name())
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ sleep(10)
+ try:
+ rc, f = write_verify_to_fs(lustrefs)
+ if not rc:
+ lutfrc(LUTF_TEST_FAIL, "Failed to verify data on " + f)
+ except:
+ return lutfrc(LUTF_TEST_FAIL, "Couldn't write the file system")
+ return lutfrc(LUTF_TEST_PASS)
+ except Exception as e:
+ raise e
+
+
--- /dev/null
+"""
+@PRIMARY: LU-10360-05
+@PRIMARY_DESC: All MGCs must have dynamic_nids set in order for the feature to work.
+@TESTCASE:
+ Setup a file system consistent of:
+ 1 node: MGS/MDS
+ 1 node: OSS should have 2 IP addresses
+ 1 node: Client
+ Mount the File System and the client
+ Use only one of the IP addresses on the OSS
+ Disable the dynamic_nids feature on both the MDS and the client
+ Write multiple files to the FS
+ Read back the files and make sure they are correct
+ unmount the OSS and remove lustre modules
+ Change the OSS configuration to use the other IP address for NID
+ load lustre on the OSS and re mount
+ LS the file system.
+ the metadata might be cached on the client and the LS could work
+ Writing to the FS should timeout and fail.
+"""
+
+import os, re, yaml, logging
+from time import sleep
+import lnetconfig
+from lutf import agents, me
+from lutf_basetest import *
+from lnet import TheLNet
+from lutf_exception import *
+from lnet_selftest import LNetSelfTest
+from lnet_helpers import LNetHelpers
+from lustre_node import SimpleLustreNode
+from lustre_roles import *
+from lustre_fs import SimpleLustreFS
+import lutf_common_def as common
+from lutf_utils import *
+
+def write_verify_to_fs(lustrefs):
+ data = generate_random_bytes(1048576)
+ files = []
+ # write 3 files
+ for i in range(0, 3):
+ files.append(lustrefs.write(0, data, thread=True))
+ # read and verify each of the files written
+ for f in files:
+ if not lustrefs.read_verify(0, data, f, thread=True):
+ return False, f
+ return True, None
+
+def run():
+ la = agents.keys()
+ if len(la) < 3:
+ return lutfrc(LUTF_TEST_SKIP,
+ msg="Not enough agents to setup a lustre FS. 3 needed %d found" % len(la))
+ try:
+ lustrefs = SimpleLustreFS(la)
+ MGSs = lustrefs.get_mgs_nodes()
+ OSSs = lustrefs.get_oss_nodes()
+ clients = lustrefs.get_client_nodes()
+ if len(MGSs) < 1 or len(OSSs) < 1 or len(clients) < 1:
+ return lutfrc(LUTF_TEST_SKIP, "Unexpected cluster: MGSs %d, OSSs %d, clients %d" % (len(MGSs), len(OSSs), len(clients)))
+
+ # configure the OSS to use the first IP address
+ oss = lustrefs.get_oss_nodes()
+ oss_intfs = oss[0].list_intfs()
+ if len(oss_intfs) <= 1:
+ return lutfrc(LUTF_TEST_SKIP, "not enough interfaces on oss %s (%s)" % (oss[0].get_node_name(), oss[0].get_node_hostname()))
+ oss_net = {'tcp': [oss_intfs[0]]}
+ lustrefs.configure_nets(oss=oss_net)
+ lustrefs.mount_servers(num_mgs=1, num_oss=1)
+ lustrefs.mount_clients(num=1)
+ rc, f = write_verify_to_fs(lustrefs)
+ if not rc:
+ lutfrc(LUTF_TEST_FAIL, "Failed to verify data on " + f)
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[1]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ sleep(10)
+ try:
+ rc, f = write_verify_to_fs(lustrefs)
+ if rc:
+ lutfrc(LUTF_TEST_FAIL, "Unexpected success")
+ except:
+ # restore the working configuration
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[0]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ # wait before returning. The intent is to give the
+ # FS sometime to recover, so the subsequent tests
+ # run
+ sleep(300)
+ return lutfrc(LUTF_TEST_PASS)
+ return lutfrc(LUTF_TEST_FAIL, "Write successful even though dynamic NIDs are disabled.")
+ except Exception as e:
+ raise e
+
+
--- /dev/null
+"""
+@PRIMARY: LU-10360-06
+@PRIMARY_DESC: All MGCs must have dynamic_nids set in order for the feature to work.
+Test dynamic_nids set on Client but not on MDT
+@TESTCASE:
+ Setup a file system consistent of:
+ 1 node: MGS/MDS
+ 1 node: OSS should have 2 IP addresses
+ 1 node: Client
+ Mount the File System and the client
+ Use only one of the IP addresses on the OSS
+ Disable the dynamic_nids feature on the MDS but enable it on the client
+ Write multiple files to the FS
+ Read back the files and make sure they are correct
+ unmount the OSS and remove lustre modules
+ Change the OSS configuration to use the other IP address for NID
+ load lustre on the OSS and re mount
+ LS the file system.
+ the metadata might be cached on the client and the LS could work
+ Writing to the FS should timeout and fail.
+"""
+
+import os, re, yaml, logging
+from time import sleep
+import lnetconfig
+from lutf import agents, me
+from lutf_basetest import *
+from lnet import TheLNet
+from lutf_exception import *
+from lnet_selftest import LNetSelfTest
+from lnet_helpers import LNetHelpers
+from lustre_node import SimpleLustreNode
+from lustre_roles import *
+from lustre_fs import SimpleLustreFS
+import lutf_common_def as common
+from lutf_utils import *
+
+def set_check_dynamic_nids(node, prev, new):
+ dn = node.get_dynamic_nids()
+ if dn != prev:
+ return False, prev
+ node.set_dynamic_nids(new)
+ dn = node.get_dynamic_nids()
+ if dn != new:
+ return False, new
+ return True, new
+
+def write_verify_to_fs(lustrefs):
+ data = generate_random_bytes(1048576)
+ files = []
+ # write 3 files
+ for i in range(0, 3):
+ files.append(lustrefs.write(0, data, thread=True))
+ # read and verify each of the files written
+ for f in files:
+ if not lustrefs.read_verify(0, data, f, thread=True):
+ return False, f
+ return True, None
+
+def run():
+ la = agents.keys()
+ if len(la) < 3:
+ return lutfrc(LUTF_TEST_SKIP,
+ msg="Not enough agents to setup a lustre FS. 3 needed %d found" % len(la))
+ try:
+ lustrefs = SimpleLustreFS(la)
+ MGSs = lustrefs.get_mgs_nodes()
+ OSSs = lustrefs.get_oss_nodes()
+ clients = lustrefs.get_client_nodes()
+ if len(MGSs) < 1 or len(OSSs) < 1 or len(clients) < 1:
+ return lutfrc(LUTF_TEST_SKIP, "Unexpected cluster: MGSs %d, OSSs %d, clients %d" % (len(MGSs), len(OSSs), len(clients)))
+
+ # configure the OSS to use the first IP address
+ oss = lustrefs.get_oss_nodes()
+ oss_intfs = oss[0].list_intfs()
+ if len(oss_intfs) <= 1:
+ return lutfrc(LUTF_TEST_SKIP, "not enough interfaces on oss %s (%s)" % (oss[0].get_node_name(), oss[0].get_node_hostname()))
+ oss_net = {'tcp': [oss_intfs[0]]}
+ lustrefs.configure_nets(oss=oss_net)
+ lustrefs.mount_servers(num_mgs=1, num_oss=1)
+ lustrefs.mount_clients(num=1)
+ rc, value = set_check_dynamic_nids(clients[0], 0, 1)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, f = write_verify_to_fs(lustrefs)
+ if not rc:
+ lutfrc(LUTF_TEST_FAIL, "Failed to verify data on " + f)
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[1]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ sleep(10)
+ try:
+ rc, f = write_verify_to_fs(lustrefs)
+ if rc:
+ lutfrc(LUTF_TEST_FAIL, "Unexpected success")
+ except:
+ # restore the working configuration
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[0]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ # wait before returning. The intent is to give the
+ # FS sometime to recover, so the subsequent tests
+ # run
+ sleep(300)
+ return lutfrc(LUTF_TEST_PASS)
+ return lutfrc(LUTF_TEST_FAIL,
+ "dynamic NID feature worked even though MDT feature is disabled.")
+ except Exception as e:
+ raise e
+
+
+
+
--- /dev/null
+"""
+@PRIMARY: LU-10360-06
+@PRIMARY_DESC: All MGCs must have dynamic_nids set in order for the feature to work.
+Test dynamic_nids set on MDT but not the client
+@TESTCASE:
+ Setup a file system consistent of:
+ 1 node: MGS/MDS
+ 1 node: OSS should have 2 IP addresses
+ 1 node: Client
+ Mount the File System and the client
+ Use only one of the IP addresses on the OSS
+ Disable the dynamic_nids feature on the client but disable it on the MDS
+ Write multiple files to the FS
+ Read back the files and make sure they are correct
+ unmount the OSS and remove lustre modules
+ Change the OSS configuration to use the other IP address for NID
+ load lustre on the OSS and re mount
+ LS the file system.
+ the metadata might be cached on the client and the LS could work
+ Writing to the FS should timeout and fail.
+"""
+
+import os, re, yaml, logging
+from time import sleep
+import lnetconfig
+from lutf import agents, me
+from lutf_basetest import *
+from lnet import TheLNet
+from lutf_exception import *
+from lnet_selftest import LNetSelfTest
+from lnet_helpers import LNetHelpers
+from lustre_node import SimpleLustreNode
+from lustre_roles import *
+from lustre_fs import SimpleLustreFS
+import lutf_common_def as common
+from lutf_utils import *
+
+def set_check_dynamic_nids(node, prev, new):
+ dn = node.get_dynamic_nids()
+ if dn != prev:
+ return False, prev
+ node.set_dynamic_nids(new)
+ dn = node.get_dynamic_nids()
+ if dn != new:
+ return False, new
+ return True, new
+
+def write_verify_to_fs(lustrefs):
+ data = generate_random_bytes(1048576)
+ files = []
+ # write 3 files
+ for i in range(0, 3):
+ files.append(lustrefs.write(0, data, thread=True))
+ # read and verify each of the files written
+ for f in files:
+ if not lustrefs.read_verify(0, data, f, thread=True):
+ return False, f
+ return True, None
+
+def run():
+ la = agents.keys()
+ if len(la) < 3:
+ return lutfrc(LUTF_TEST_SKIP,
+ msg="Not enough agents to setup a lustre FS. 3 needed %d found" % len(la))
+ try:
+ lustrefs = SimpleLustreFS(la)
+ MGSs = lustrefs.get_mgs_nodes()
+ OSSs = lustrefs.get_oss_nodes()
+ clients = lustrefs.get_client_nodes()
+ if len(MGSs) < 1 or len(OSSs) < 1 or len(clients) < 1:
+ return lutfrc(LUTF_TEST_SKIP, "Unexpected cluster: MGSs %d, OSSs %d, clients %d" % (len(MGSs), len(OSSs), len(clients)))
+
+ # configure the OSS to use the first IP address
+ oss = lustrefs.get_oss_nodes()
+ oss_intfs = oss[0].list_intfs()
+ if len(oss_intfs) <= 1:
+ return lutfrc(LUTF_TEST_SKIP, "not enough interfaces on oss %s (%s)" % (oss[0].get_node_name(), oss[0].get_node_hostname()))
+ oss_net = {'tcp': [oss_intfs[0]]}
+ lustrefs.configure_nets(oss=oss_net)
+ lustrefs.mount_servers(num_mgs=1, num_oss=1)
+ lustrefs.mount_clients(num=1)
+ rc, value = set_check_dynamic_nids(MGSs[0], 0, 1)
+ if not rc:
+ return lutfrc(LUTF_TEST_FAIL, "expected dynamic_nids to be set to: " + str(value))
+ rc, f = write_verify_to_fs(lustrefs)
+ if not rc:
+ lutfrc(LUTF_TEST_FAIL, "Failed to verify data on " + f)
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[1]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ sleep(10)
+ try:
+ rc, f = write_verify_to_fs(lustrefs)
+ if rc:
+ lutfrc(LUTF_TEST_FAIL, "Unexpected success")
+ except:
+ # restore the working configuration
+ oss[0].unconfigure_lustre()
+ oss_net = {'tcp': [oss_intfs[0]]}
+ oss[0].configure_net(oss_net)
+ oss[0].commit()
+ oss[0].configure_lustre(lustrefs.get_mgs_nids())
+ # wait before returning. The intent is to give the
+ # FS sometime to recover, so the subsequent tests
+ # run
+ sleep(300)
+ return lutfrc(LUTF_TEST_PASS)
+ return lutfrc(LUTF_TEST_FAIL,
+ "dynamic NID feature worked even though MDT feature is disabled.")
+ except Exception as e:
+ raise e
+
+
+
+
git://git.whamcloud.com / fs / lustre-release.git / commitdiff