def test_import_gluster():
tendrlNS = init()
# import pdb; pdb.set_trace();
NS.compiled_definitions = tendrlNS.current_ns.definitions
parameters = maps.NamedDict(job_id=1, flow_id=1)
ret_val, err = gluster_help.import_gluster(parameters)
assert ret_val is False
assert err is not None
NS.config.data['package_source_type'] = 'pip'
with patch.object(ansible_module_runner.AnsibleRunner, 'run', run):
ret, err = gluster_help.import_gluster(parameters)
assert ret is False
with patch.object(ansible_module_runner.AnsibleRunner, 'run',
return_value=({"rc": 1, "msg": None}, None)):
ret, err = gluster_help.import_gluster(parameters)
assert ret is False
NS.config.data['package_source_type'] = 'rpm'
with patch.object(ansible_module_runner.AnsibleRunner, 'run',
return_value=({"rc": 0, "msg": None}, None)):
with patch.object(__builtin__, 'open', open):
with patch.object(cmd_utils.Command, 'run',
return_value=("err", "", 1)):
ret, err = gluster_help.import_gluster(parameters)
assert ret is False
with patch.object(ansible_module_runner.AnsibleRunner, 'run',
return_value=({"rc": 0, "msg": None}, None)):
with patch.object(__builtin__, 'open', open):
with patch.object(cmd_utils.Command, 'run',
return_value=(None, "", 0)):
with patch.object(ServiceStatus, 'status',
return_value=True):
ret, err = gluster_help.import_gluster(parameters)
assert ret is True
def test_import_gluster():
tendrlNS = init()
# import pdb; pdb.set_trace();
NS.compiled_definitions = tendrlNS.current_ns.definitions
parameters = maps.NamedDict(job_id=1, flow_id=1)
ret_val, err = gluster_help.import_gluster(parameters)
assert ret_val is False
assert err is not None
NS.config.data['package_source_type'] = 'pip'
with patch.object(ansible_module_runner.AnsibleRunner, 'run', run):
ret, err = gluster_help.import_gluster(parameters)
assert ret is False
with patch.object(ansible_module_runner.AnsibleRunner,
'run',
return_value=({
"rc": 1,
"msg": None
}, None)):
ret, err = gluster_help.import_gluster(parameters)
assert ret is False
NS.config.data['package_source_type'] = 'rpm'
with patch.object(ansible_module_runner.AnsibleRunner,
'run',
return_value=({
"rc": 0,
"msg": None
}, None)):
with patch.object(__builtin__, 'open', open):
with patch.object(cmd_utils.Command,
'run',
return_value=("err", "", 1)):
ret, err = gluster_help.import_gluster(parameters)
assert ret is False
with patch.object(ansible_module_runner.AnsibleRunner,
'run',
return_value=({
"rc": 0,
"msg": None
}, None)):
with patch.object(__builtin__, 'open', open):
with patch.object(cmd_utils.Command,
'run',
return_value=(None, "", 0)):
with patch.object(ServiceStatus, 'status', return_value=True):
ret, err = gluster_help.import_gluster(parameters)
assert ret is True
def test_import_gluster():
tendrlNS = init()
NS.compiled_definitions = tendrlNS.current_ns.definitions
parameters = maps.NamedDict(job_id=1, flow_id=1)
assert gluster_help.import_gluster(parameters) is False
NS.config.data['package_source_type'] = 'pip'
with patch.object(ansible_module_runner.AnsibleRunner, 'run',
run) as mock_run:
ret = gluster_help.import_gluster(parameters)
assert ret is False
with patch.object(ansible_module_runner.AnsibleRunner,
'run',
return_value=True) as mock_run:
with patch.object(__builtin__, 'open', open) as mock_open:
ret = gluster_help.import_gluster(parameters)
with patch.object(ansible_module_runner, 'AnsibleRunner',
ansible) as mock_ansible:
with pytest.raises(ansible_module_runner.AnsibleModuleNotFound):
ret = gluster_help.import_gluster(parameters)
NS.config.data['package_source_type'] = 'rpm'
with patch.object(ansible_module_runner.AnsibleRunner, 'run',
run) as mock_run:
ret = gluster_help.import_gluster(parameters)
assert ret is False
def run(self):
try:
# Lock nodes
create_cluster_utils.acquire_node_lock(self.parameters)
integration_id = self.parameters['TendrlContext.integration_id']
sds_name = self.parameters['DetectedCluster.sds_pkg_name']
if not self.parameters.get('import_after_expand', False) and \
not self.parameters.get('import_after_create', False):
# check if gdeploy in already provisioned in this cluster
# if no it has to be provisioned here
if sds_name.find("gluster") > -1 and \
not self.parameters.get("gdeploy_provisioned", False) and \
not self._probe_and_mark_provisioner(
self.parameters["Node[]"], integration_id
):
create_cluster_utils.install_gdeploy()
create_cluster_utils.install_python_gdeploy()
ssh_job_ids = create_cluster_utils.gluster_create_ssh_setup_jobs(
self.parameters)
while True:
gevent.sleep(3)
all_status = {}
for job_id in ssh_job_ids:
all_status[job_id] = NS._int.client.read(
"/queue/%s/status" % job_id).value
_failed = {
_jid: status
for _jid, status in all_status.iteritems()
if status == "failed"
}
if _failed:
raise AtomExecutionFailedError(
"SSH setup failed for jobs %s cluster %s" %
(str(_failed), integration_id))
if all([
status == "finished"
for status in all_status.values()
]):
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"SSH setup completed for all nodes in cluster %s"
% integration_id
}))
# set this node as gluster provisioner
tags = ["provisioner/%s" % integration_id]
NS.node_context = NS.node_context.load()
tags += NS.node_context.tags
NS.node_context.tags = list(set(tags))
NS.node_context.save()
# set gdeploy_provisioned to true so that no other nodes
# tries to configure gdeploy
self.parameters['gdeploy_provisioned'] = True
break
NS.tendrl_context = NS.tendrl_context.load()
NS.tendrl_context.integration_id = integration_id
_detected_cluster = NS.tendrl.objects.DetectedCluster().load()
NS.tendrl_context.cluster_id = _detected_cluster.detected_cluster_id
NS.tendrl_context.cluster_name = _detected_cluster.detected_cluster_name
NS.tendrl_context.sds_name = _detected_cluster.sds_pkg_name
NS.tendrl_context.sds_version = _detected_cluster.sds_pkg_version
NS.tendrl_context.save()
Event(
Message(job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Registered Node %s with cluster %s" %
(NS.node_context.node_id,
NS.tendrl_context.integration_id)
}))
node_list = self.parameters['Node[]']
cluster_nodes = []
if len(node_list) > 1:
# This is the master node for this flow
for node in node_list:
if NS.node_context.node_id != node:
new_params = self.parameters.copy()
new_params['Node[]'] = [node]
# create same flow for each node in node list except $this
payload = {
"tags": ["tendrl/node_%s" % node],
"run": "tendrl.flows.ImportCluster",
"status": "new",
"parameters": new_params,
"parent": self.parameters['job_id'],
"type": "node"
}
_job_id = str(uuid.uuid4())
cluster_nodes.append(_job_id)
Job(job_id=_job_id, status="new",
payload=payload).save()
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Importing (job: %s) Node %s to cluster %s"
% (_job_id, node, integration_id)
}))
if "ceph" in sds_name.lower():
node_context = NS.node_context.load()
is_mon = False
for tag in node_context.tags:
mon_tag = NS.compiled_definitions.get_parsed_defs(
)['namespace.tendrl']['tags']['ceph-mon']
if mon_tag in tag:
is_mon = True
if is_mon:
# Check if minimum required version of underlying ceph
# cluster met. If not fail the import task
detected_cluster = NS.tendrl.objects.DetectedCluster(
).load()
detected_cluster_ver = detected_cluster.sds_pkg_version.split(
'.')
maj_ver = detected_cluster_ver[0]
min_ver = detected_cluster_ver[1]
reqd_ceph_ver = NS.compiled_definitions.get_parsed_defs(
)['namespace.tendrl']['min_reqd_ceph_ver']
req_maj_ver, req_min_ver, req_rel = reqd_ceph_ver.split(
'.')
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Check: Minimum required version (%s.%s.%s) of Ceph Storage"
% (req_maj_ver, req_min_ver, req_rel)
}))
if int(maj_ver) < int(req_maj_ver) or \
int(min_ver) < int(req_min_ver):
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="error",
publisher=NS.publisher_id,
payload={
"message":
"Error: Minimum required version (%s.%s.%s) "
"doesnt match that of detected Ceph Storage (%s.%s.%s)"
% (req_maj_ver, req_min_ver, req_rel,
maj_ver, min_ver, 0)
}))
raise FlowExecutionFailedError(
"Detected ceph version: %s"
" is lesser than required version: %s" %
(detected_cluster.sds_pkg_version, reqd_ceph_ver))
import_ceph(self.parameters)
else:
# Check if minimum required version of underlying gluster
# cluster met. If not fail the import task
detected_cluster = NS.tendrl.objects.DetectedCluster().load()
detected_cluster_ver = detected_cluster.sds_pkg_version.split(
'.')
maj_ver = detected_cluster_ver[0]
min_ver = detected_cluster_ver[1]
reqd_gluster_ver = NS.compiled_definitions.get_parsed_defs(
)['namespace.tendrl']['min_reqd_gluster_ver']
req_maj_ver, req_min_ver, req_rel = reqd_gluster_ver.split('.')
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Check: Minimum required version (%s.%s.%s) of Gluster Storage"
% (req_maj_ver, req_min_ver, req_rel)
}))
if int(maj_ver) < int(req_maj_ver) or \
int(min_ver) < int(req_min_ver):
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="error",
publisher=NS.publisher_id,
payload={
"message":
"Error: Minimum required version (%s.%s.%s) "
"doesnt match that of detected Gluster Storage (%s.%s.%s)"
% (req_maj_ver, req_min_ver, req_rel, maj_ver,
min_ver, 0)
}))
raise FlowExecutionFailedError(
"Detected gluster version: %s"
" is lesser than required version: %s" %
(detected_cluster.sds_pkg_version, reqd_gluster_ver))
import_gluster(self.parameters)
Event(
Message(job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Waiting for participant nodes %s to be "
"imported %s" % (node_list, integration_id)
}))
# An import is sucessfull once all Node[] register to
# /clusters/:integration_id/nodes/:node_id
while True:
_all_node_status = []
gevent.sleep(3)
for node_id in self.parameters['Node[]']:
_status = NS.tendrl.objects.ClusterNodeContext(node_id=node_id).exists() \
and NS.tendrl.objects.ClusterTendrlContext(
integration_id=integration_id
).exists()
_all_node_status.append(_status)
if _all_node_status:
if all(_all_node_status):
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Import Cluster completed for all nodes "
"in cluster %s" % integration_id
}))
break
Event(
Message(job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message":
"Sucessfully imported cluster %s" % integration_id
}))
except Exception as ex:
# For traceback
Event(
ExceptionMessage(priority="error",
publisher=NS.publisher_id,
payload={
"message": ex.message,
"exception": ex
}))
# raising exception to mark job as failed
raise ex
finally:
# release lock
create_cluster_utils.release_node_lock(self.parameters)
return True
def run(self):
try:
integration_id = self.parameters['TendrlContext.integration_id']
_cluster = NS.tendrl.objects.Cluster(
integration_id=integration_id
).load()
# Lock nodes
flow_utils.acquire_node_lock(self.parameters)
NS.tendrl_context = NS.tendrl_context.load()
# TODO(team) when Tendrl supports create/expand/shrink cluster
# setup passwordless ssh for all gluster nodes with given
# integration_id (check
# /indexes/tags/tendrl/integration/$integration_id for list of
# nodes in cluster
node_list = self.parameters['Node[]']
cluster_nodes = []
if len(node_list) > 1:
# This is the master node for this flow
for node in node_list:
if NS.node_context.node_id != node:
new_params = self.parameters.copy()
new_params['Node[]'] = [node]
# create same flow for each node in node list except
# $this
payload = {"tags": ["tendrl/node_%s" % node],
"run": "tendrl.flows.ImportCluster",
"status": "new",
"parameters": new_params,
"parent": self.parameters['job_id'],
"type": "node"
}
_job_id = str(uuid.uuid4())
cluster_nodes.append(_job_id)
NS.tendrl.objects.Job(
job_id=_job_id,
status="new",
payload=payload
).save()
logger.log(
"info",
NS.publisher_id,
{"message": "ImportCluster %s (jobID: %s) :"
"importing host %s" %
(_cluster.short_name, _job_id, node)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
# Check if minimum required version of underlying gluster
# cluster met. If not fail the import task
# A sample output from "rpm -qa | grep glusterfs-server"
# looks as below
# `glusterfs-server-3.8.4-54.4.el7rhgs.x86_64`
# In case of upstream build the format could be as below
# `glusterfs-server-4.1dev-0.203.gitc3e1a2e.el7.centos.x86_64`
# `glusterfs-server-3.12.8-0.0.el7.centos.x86_64.rpm`
cmd = subprocess.Popen(
'rpm -q glusterfs-server',
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
)
out, err = cmd.communicate()
if out in [None, ""] or err:
logger.log(
"error",
NS.publisher_id,
{"message": "Failed to detect underlying cluster version"},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
lines = out.split('\n')
build_no = None
req_build_no = None
ver_det = lines[0].split('glusterfs-server-')[-1].split('.')
maj_ver = ver_det[0]
min_ver = ver_det[1]
if 'dev' in min_ver:
min_ver = min_ver[0]
rel = ver_det[2]
if '-' in rel:
build_no = rel.split('-')[-1]
rel = rel.split('-')[0]
reqd_gluster_ver = NS.compiled_definitions.get_parsed_defs()[
'namespace.tendrl'
]['min_reqd_gluster_ver']
req_maj_ver, req_min_ver, req_rel = reqd_gluster_ver.split('.')
if '-' in req_rel:
req_build_no = req_rel.split('-')[-1]
req_rel = req_rel.split('-')[0]
logger.log(
"info",
NS.publisher_id,
{"message": "Checking minimum required version ("
"%s.%s.%s) of Gluster Storage" %
(req_maj_ver, req_min_ver, req_rel)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
ver_check_failed = False
if int(maj_ver) < int(req_maj_ver):
ver_check_failed = True
else:
if int(maj_ver) == int(req_maj_ver):
if int(min_ver) < int(req_min_ver):
ver_check_failed = True
else:
if int(min_ver) == int(req_min_ver):
if int(rel) < int(req_rel):
ver_check_failed = True
else:
if int(rel) == int(req_rel):
if build_no is not None and \
req_build_no is not None and \
int(build_no) < int(req_build_no):
ver_check_failed = True
if ver_check_failed:
logger.log(
"error",
NS.publisher_id,
{"message": "Error: Minimum required version "
"(%s.%s.%s) "
"doesnt match that of detected Gluster "
"Storage (%s.%s.%s)" %
(req_maj_ver, req_min_ver, req_rel,
maj_ver, min_ver, 0)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
ret_val, err = import_gluster(self.parameters)
if not ret_val:
logger.log(
"error",
NS.publisher_id,
{"message": "Error importing the cluster (integration_id:"
" %s). Error: %s" % (integration_id, err)
},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
if len(node_list) > 1:
logger.log(
"info",
NS.publisher_id,
{"message": "ImportCluster %s waiting for hosts %s "
"to be imported" % (_cluster.short_name, node_list)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
loop_count = 0
# Wait for (no of nodes) * 6 minutes for import to complete
wait_count = (len(node_list) - 1) * 36
while True:
child_jobs_failed = []
parent_job = NS.tendrl.objects.Job(
job_id=self.parameters['job_id']
).load()
if loop_count >= wait_count:
logger.log(
"error",
NS.publisher_id,
{"message": "Import jobs on cluster(%s) not yet "
"complete on all nodes(%s). Timing out." %
(_cluster.short_name, str(node_list))},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
# Marking child jobs as failed which did not complete
# as the parent job has timed out. This has to be done
# explicitly because these jobs will still be processed
# by the node-agent, and will keep it busy, which might
# defer the new jobs or lead to their timeout.
for child_job_id in parent_job.children:
child_job = NS.tendrl.objects.Job(
job_id=child_job_id
).load()
if child_job.status not in ["finished", "failed"]:
child_job.status = "failed"
child_job.save()
return False
time.sleep(10)
completed = True
for child_job_id in parent_job.children:
child_job = NS.tendrl.objects.Job(
job_id=child_job_id
).load()
if child_job.status not in ["finished", "failed"]:
completed = False
elif child_job.status == "failed":
child_jobs_failed.append(child_job.job_id)
if completed:
break
else:
loop_count += 1
continue
if len(child_jobs_failed) > 0:
_msg = "Child jobs failed are %s" % child_jobs_failed
logger.log(
"error",
NS.publisher_id,
{"message": _msg},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
except Exception as ex:
# For traceback
Event(
ExceptionMessage(
priority="error",
publisher=NS.publisher_id,
payload={
"message": ex.message,
"exception": ex
}
)
)
# raising exception to mark job as failed
raise ex
finally:
# release lock
flow_utils.release_node_lock(self.parameters)
return True
def run(self):
try:
integration_id = self.parameters['TendrlContext.integration_id']
_cluster = NS.tendrl.objects.Cluster(
integration_id=integration_id
).load()
# Lock nodes
flow_utils.acquire_node_lock(self.parameters)
NS.tendrl_context = NS.tendrl_context.load()
# TODO(team) when Tendrl supports create/expand/shrink cluster
# setup passwordless ssh for all gluster nodes with given
# integration_id (check
# /indexes/tags/tendrl/integration/$integration_id for list of
# nodes in cluster
node_list = self.parameters['Node[]']
cluster_nodes = []
if len(node_list) > 1:
# This is the master node for this flow
# Find number of volumes in the cluster to run profiling job
cmd = cmd_utils.Command('gluster volume list')
out, err, rc = cmd.run()
if not err:
volumes = filter(None, out.split("\n"))
ret_val, err = enable_disable_volume_profiling(
volumes, self.parameters)
if not ret_val:
logger.log(
"error",
NS.publisher_id,
{"message": "Failed to %s profiling. Error: %s"
% (_cluster.volume_profiling_flag, err)
},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
else:
logger.log(
"warning",
NS.publisher_id,
{"message": "Unable to find list of volumes in a "
"cluster, Skipping volume enable/disable "
"profiling flow. error: %s" % err},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
# continue the import flow
for node in node_list:
if NS.node_context.node_id != node:
new_params = self.parameters.copy()
new_params['Node[]'] = [node]
# create same flow for each node in node list except
# $this
payload = {"tags": ["tendrl/node_%s" % node],
"run": "tendrl.flows.ImportCluster",
"status": "new",
"parameters": new_params,
"parent": self.parameters['job_id'],
"type": "node"
}
_job_id = str(uuid.uuid4())
cluster_nodes.append(_job_id)
NS.tendrl.objects.Job(
job_id=_job_id,
status="new",
payload=payload
).save()
logger.log(
"info",
NS.publisher_id,
{"message": "ImportCluster %s (jobID: %s) :"
"importing host %s" %
(_cluster.short_name, _job_id, node)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
# Check if minimum required version of underlying gluster
# cluster met. If not fail the import task
# A sample output from "rpm -qa | grep glusterfs-server"
# looks as below
# `glusterfs-server-3.8.4-54.4.el7rhgs.x86_64`
# In case of upstream build the format could be as below
# `glusterfs-server-4.1dev-0.203.gitc3e1a2e.el7.centos.x86_64`
# `glusterfs-server-3.12.8-0.0.el7.centos.x86_64.rpm`
cmd = subprocess.Popen(
'rpm -q glusterfs-server',
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
)
out, err = cmd.communicate()
if out in [None, ""] or err:
logger.log(
"error",
NS.publisher_id,
{"message": "Failed to detect underlying cluster version"},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
lines = out.split('\n')
build_no = None
req_build_no = None
ver_det = lines[0].split('glusterfs-server-')[-1].split('.')
maj_ver = ver_det[0]
min_ver = ver_det[1]
if 'dev' in min_ver:
min_ver = min_ver[0]
rel = ver_det[2]
if '-' in rel:
build_no = rel.split('-')[-1]
rel = rel.split('-')[0]
reqd_gluster_ver = NS.compiled_definitions.get_parsed_defs()[
'namespace.tendrl'
]['min_reqd_gluster_ver']
req_maj_ver, req_min_ver, req_rel = reqd_gluster_ver.split('.')
if '-' in req_rel:
req_build_no = req_rel.split('-')[-1]
req_rel = req_rel.split('-')[0]
logger.log(
"info",
NS.publisher_id,
{"message": "Checking minimum required version ("
"%s.%s.%s) of Gluster Storage" %
(req_maj_ver, req_min_ver, req_rel)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
ver_check_failed = False
if int(maj_ver) < int(req_maj_ver):
ver_check_failed = True
else:
if int(maj_ver) == int(req_maj_ver):
if int(min_ver) < int(req_min_ver):
ver_check_failed = True
else:
if int(min_ver) == int(req_min_ver):
if int(rel) < int(req_rel):
ver_check_failed = True
else:
if int(rel) == int(req_rel):
if build_no is not None and \
req_build_no is not None and \
int(build_no) < int(req_build_no):
ver_check_failed = True
if ver_check_failed:
logger.log(
"error",
NS.publisher_id,
{"message": "Error: Minimum required version "
"(%s.%s.%s) "
"doesnt match that of detected Gluster "
"Storage (%s.%s.%s)" %
(req_maj_ver, req_min_ver, req_rel,
maj_ver, min_ver, 0)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
ret_val, err = import_gluster(self.parameters)
if not ret_val:
logger.log(
"error",
NS.publisher_id,
{"message": "Error importing the cluster (integration_id:"
" %s). Error: %s" % (integration_id, err)
},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
if len(node_list) > 1:
# find number of volumes in a cluster
cmd = cmd_utils.Command('gluster volume list')
out, err, rc = cmd.run()
# default intervel is 8 min
# 5 sec sleep for one count increment (480 / 5)
wait_count = 96
if not err:
volumes = out.split("\n")
# 15 sec for each volume
wait_count = wait_count + (len(volumes) * 3)
logger.log(
"info",
NS.publisher_id,
{"message": "ImportCluster %s waiting for hosts %s "
"to be imported" % (_cluster.short_name, node_list)},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
loop_count = 0
# Wait for (no of nodes) * 6 minutes for import to complete
while True:
child_jobs_failed = []
parent_job = NS.tendrl.objects.Job(
job_id=self.parameters['job_id']
).load()
if loop_count >= wait_count:
logger.log(
"error",
NS.publisher_id,
{"message": "Import jobs on cluster(%s) not yet "
"complete on all nodes(%s). Timing out." %
(_cluster.short_name, str(node_list))},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
# Marking child jobs as failed which did not complete
# as the parent job has timed out. This has to be done
# explicitly because these jobs will still be processed
# by the node-agent, and will keep it busy, which might
# defer the new jobs or lead to their timeout.
for child_job_id in parent_job.children:
child_job = NS.tendrl.objects.Job(
job_id=child_job_id
).load()
if child_job.status not in ["finished", "failed"]:
child_job.status = "failed"
child_job.save()
return False
time.sleep(10)
completed = True
for child_job_id in parent_job.children:
child_job = NS.tendrl.objects.Job(
job_id=child_job_id
).load()
if child_job.status not in ["finished", "failed"]:
completed = False
elif child_job.status == "failed":
child_jobs_failed.append(child_job.job_id)
if completed:
break
else:
loop_count += 1
continue
if len(child_jobs_failed) > 0:
_msg = "Child jobs failed are %s" % child_jobs_failed
logger.log(
"error",
NS.publisher_id,
{"message": _msg},
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id']
)
return False
except Exception as ex:
# For traceback
Event(
ExceptionMessage(
priority="error",
publisher=NS.publisher_id,
payload={
"message": ex.message,
"exception": ex
}
)
)
# raising exception to mark job as failed
raise ex
finally:
# release lock
flow_utils.release_node_lock(self.parameters)
return True
def run(self):
try:
integration_id = self.parameters['TendrlContext.integration_id']
# Lock nodes
create_cluster_utils.acquire_node_lock(self.parameters)
NS.tendrl_context = NS.tendrl_context.load()
# TODO(team) when Tendrl supports create/expand/shrink cluster
# setup passwordless ssh for all gluster nodes with given
# integration_id (check
# /indexes/tags/tendrl/integration/$integration_id for list of
# nodes in cluster
node_list = self.parameters['Node[]']
cluster_nodes = []
if len(node_list) > 1:
# This is the master node for this flow
for node in node_list:
if NS.node_context.node_id != node:
new_params = self.parameters.copy()
new_params['Node[]'] = [node]
# create same flow for each node in node list except
# $this
payload = {"tags": ["tendrl/node_%s" % node],
"run": "tendrl.flows.ImportCluster",
"status": "new",
"parameters": new_params,
"parent": self.parameters['job_id'],
"type": "node"
}
_job_id = str(uuid.uuid4())
cluster_nodes.append(_job_id)
Job(job_id=_job_id,
status="new",
payload=payload).save()
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message": "Importing (job: %s) Node %s "
"to cluster %s" %
(_job_id, node, integration_id)
}
)
)
# Check if minimum required version of underlying gluster
# cluster met. If not fail the import task
cluster_ver = \
NS.tendrl_context.sds_version.split('.')
maj_ver = cluster_ver[0]
min_ver = re.findall(r'\d+', cluster_ver[1])[0]
reqd_gluster_ver = NS.compiled_definitions.get_parsed_defs()[
'namespace.tendrl'
]['min_reqd_gluster_ver']
req_maj_ver, req_min_ver, req_rel = reqd_gluster_ver.split('.')
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message": "Check: Minimum required version ("
"%s.%s.%s) of Gluster Storage" %
(req_maj_ver, req_min_ver, req_rel)
}
)
)
ver_check_failed = False
if int(maj_ver) < int(req_maj_ver):
ver_check_failed = True
else:
if int(maj_ver) == int(req_maj_ver) and \
int(min_ver) < int(req_min_ver):
ver_check_failed = True
if ver_check_failed:
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="error",
publisher=NS.publisher_id,
payload={
"message": "Error: Minimum required version "
"(%s.%s.%s) "
"doesnt match that of detected Gluster "
"Storage (%s.%s.%s)" %
(req_maj_ver, req_min_ver, req_rel,
maj_ver, min_ver, 0)
}
)
)
raise FlowExecutionFailedError(
"Detected gluster version: %s"
" is lesser than required version: %s" %
(
NS.tendrl_context.sds_version,
reqd_gluster_ver
)
)
if not import_gluster(self.parameters):
return False
if len(node_list) > 1:
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message": "Waiting for participant nodes %s to "
"be "
"imported %s" % (node_list, integration_id)
}
)
)
loop_count = 0
# Wait for (no of nodes) * 6 minutes for import to complete
wait_count = (len(node_list) - 1) * 36
while True:
parent_job = Job(job_id=self.parameters['job_id']).load()
if loop_count >= wait_count:
Event(
Message(
job_id=self.parameters['job_id'],
flow_id=self.parameters['flow_id'],
priority="info",
publisher=NS.publisher_id,
payload={
"message": "Import jobs not yet complete "
"on all nodes. Timing out. (%s, %s)" %
(str(node_list), integration_id)
}
)
)
return False
time.sleep(10)
finished = True
for child_job_id in parent_job.children:
child_job = Job(job_id=child_job_id).load()
if child_job.status != "finished":
finished = False
break
if finished:
break
else:
loop_count += 1
continue
except Exception as ex:
# For traceback
Event(
ExceptionMessage(
priority="error",
publisher=NS.publisher_id,
payload={
"message": ex.message,
"exception": ex
}
)
)
# raising exception to mark job as failed
raise ex
finally:
# release lock
create_cluster_utils.release_node_lock(self.parameters)
return True
