def delete_all_pods_for_service(dto):
"""
停止所有的某类容器[根据关键字删]
:param service: 容器名中的关键字
:return: 结果
"""
service = dto["service"]
result_list = []
target_host = "10.60.38.181"
name_list = K8sObserver.get_pod_name_list('sock-shop')
for pod_name in name_list:
if service in pod_name and (service + "-db") not in pod_name and (
service + "-external") not in pod_name:
target_inject = Cmd['k8s'] + pod_name
r = Runner()
r.run_ad_hoc(hosts=target_host,
module='shell',
args=target_inject)
result = r.get_adhoc_result()
result_list.append(
handle_inject_result('k8s', target_host, target_inject,
result,
sys._getframe().f_code.co_name,
dto['open']))
return result_list
def chaos_inject_mem(dto):
find = 0
timeout = ''
(target_host, is_exist) = get_target_host(dto)
if not is_exist:
return 'Host: ' + target_host + ' does not exist.'
if dto['percent'] == 'default':
target_inject = Default_cmd['mem']
else:
target_inject = Cmd['mem'] + dto['percent']
if dto['timeout'] == 'default':
timeout = ' --timeout 300'
elif dto['timeout'] != 'no':
timeout = ' --timeout ' + dto['timeout']
for i in range(0, len(has_injected)):
if has_injected[i]['host'] == target_host and has_injected[i][
'inject_type'] == 'mem':
find = 1
if find == 0:
r = Runner()
r.run_ad_hoc(hosts=target_host,
module='shell',
args=target_inject + timeout)
result = r.get_adhoc_result()
return handle_inject_result('mem', target_host,
target_inject + timeout, result,
sys._getframe().f_code.co_name,
dto['open'])
else:
Logger.log(
"error", "HOST HAS BEEN INJECTED - Method : " +
sys._getframe().f_code.co_name + "() - - " + target_host)
return "The host's mem has been injected"
def chaos_inject_network(dto):
find = 0
time = '3000'
timeout = '600'
(target_host, is_exist) = get_target_host(dto)
if not is_exist:
return 'Host: ' + target_host + ' does not exist.'
if dto['time'] != 'default':
time = dto['time']
if dto['timeout'] != 'default':
timeout = dto['timeout']
target_inject = Cmd[
'network'] + '--time ' + time + ' --timeout ' + timeout
for i in range(0, len(has_injected)):
if has_injected[i]['host'] == target_host and has_injected[i][
'inject_type'] == 'network':
find = 1
if find == 0:
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=target_inject)
result = r.get_adhoc_result()
return handle_inject_result('network', target_host,
target_inject + timeout, result,
sys._getframe().f_code.co_name,
dto['open'])
else:
return "The host's network has been injected"
def inject_io(dto):
r = Runner()
r.run_ad_hoc(hosts=dto['host'],
module='shell',
args='stress -i 100 -t ' + dto['inject_duration'] +
' > /dev/null 2>&1')
result = r.get_adhoc_result()
return result
def inject_mem(dto):
r = Runner()
r.run_ad_hoc(hosts=dto['host'],
module='shell',
args='stress --vm 4 --vm-bytes 1G --vm-hang ' +
dto['inject_duration'] + ' -t ' + dto['inject_duration'] +
' > /dev/null 2>&1')
result = r.get_adhoc_result()
return result
def chaosinjectvm(dto):
i = random.randint(0, 4)
j = random.randint(0, 2)
r = Runner()
r.run_ad_hoc(hosts=HostsVM[i],
module='shell',
args=Cmd[j] + " --timeout " + dto['inject_duration'])
result = r.get_adhoc_result()
return result
def chaosinject1(dto):
#i=random.randint(0, 4)
#j=random.randint(0, 2)
r = Runner()
r.run_ad_hoc(hosts="192.168.199.31",
module='shell',
args="blade create disk fill --size 1000 --timeout " +
dto['inject_duration'])
result = r.get_adhoc_result()
return result
def view_inject_on_host_by_status(dto):
(target_host, is_exist) = get_target_host(dto)
if not is_exist:
return 'Host: ' + dto['host'] + ' does not exist.'
target_inject = "./blade status --type create"
status_type = str(dto['status']).capitalize()
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=target_inject)
result = r.get_adhoc_result()
inject_list = []
if len(result['success']) > 0:
transform_ip = result['success'].keys()[0]
info = \
json.loads(result['success'][transform_ip]['stdout'].encode('unicode-escape').decode('string_escape'))[
'result']
for i in info:
if i['Status'] == status_type:
inject_list.append(i)
try:
requests.post(url=target_url,
params={'content': str(inject_list)},
verify=False,
timeout=2)
except Timeout:
pass
finally:
Logger.log(
"info", "SUCCESS - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(dto))
return inject_list
else:
if len(result["unreachable"]) > 0:
transform_ip = result["unreachable"].keys()[0]
message = result["unreachable"][transform_ip]["msg"]
flag = "UNREACHABLE"
else:
transform_ip = result["failed"].keys()[0]
message = result["failed"][transform_ip]["msg"]
flag = "FAILED"
view_info = {
"ip": target_host,
"status type": status_type,
"cmd": target_inject,
"message": message
}
Logger.log(
"error", flag + " - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(view_info))
return inject_list
def stop_all_chaos_inject_on_all_nodes(mq_control):
uid_list = []
result_list = []
for target_host in Hosts:
dto = {
'host': target_host,
}
target_inject = './blade status --type create'
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=target_inject)
result = r.get_adhoc_result()
if len(result["success"]) > 0:
transform_ip = result["success"].keys()[0]
info = json.loads(
result["success"][transform_ip]["stdout"].encode(
'unicode-escape').decode('string_escape'))["result"]
for i in info:
if i["Status"] == "Success":
uid_list.append(i["Uid"])
if len(uid_list) == 0:
Logger.log(
'info', 'NO CHAOS INJECT - Method : ' +
sys._getframe().f_code.co_name + "() - - " + str(dto))
for item in uid_list:
cmd = './blade destroy ' + item
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=cmd)
result = r.get_adhoc_result()
if len(result["success"]) > 0:
global inject_info
global has_injected
inject_info = []
has_injected = []
# for i in range(0, len(inject_info)):
# if inject_info[i]['cmd_id'] == item:
# target_host = inject_info[i]['ip']
# inject_info.pop(i)
# for i in range(0, len(has_injected)):
# if target_host == has_injected[i]['host'] and item == has_injected[i]['tag']:
# has_injected.pop(i)
# break
result_list.append(
handle_inject_result("destroy", target_host, cmd, result,
sys._getframe().f_code.co_name,
mq_control['open']))
return result_list
def get_pod_log(pod_name, namespace, request_time):
"""
获取某个namespace下, 指定pod的日志, 并输出到指定的文件夹
:param pod_name: pod name
:param namespace: namespace
:param request_time: 发出请求的时间
:return: success or failed
"""
dir_name = request_time
command = '/opt/kube/bin/kubectl logs ' + pod_name + ' -n ' + namespace
r = Runner()
r.run_ad_hoc(hosts="10.60.38.181", module='shell', args=command)
result = r.get_adhoc_result()
if len(result["success"]) > 0:
transform_ip = result["success"].keys()[0]
stdout_lines = result["success"][transform_ip]["stdout_lines"]
cwd = os.getcwd()
dir_path = cwd + "/pod_log_dir/" + str(dir_name)
is_exist = os.path.exists(dir_path)
if not is_exist:
os.makedirs(dir_path)
log_path = dir_path + "/" + pod_name + ".txt"
with open(log_path, 'a') as f:
for line in stdout_lines:
f.write(line)
f.write('\n')
info = {"pod name": pod_name, "command": command, "log path": log_path}
Logger.log(
"info", "SUCCESS - Method : " + sys._getframe().f_code.co_name +
"() - - " + str(info))
else:
if len(result["unreachable"]) > 0:
transform_ip = result["unreachable"].keys()[0]
message = result["unreachable"][transform_ip]["msg"]
flag = "UNREACHABLE"
else:
transform_ip = result["failed"].keys()[0]
message = result["failed"][transform_ip]["msg"]
flag = "FAILED"
info = {"pod name": pod_name, "command": command, "message": message}
Logger.log(
"error", flag + " - Method : " + sys._getframe().f_code.co_name +
"() - - " + str(info))
return info
def chaos_inject_random(dto):
global Default_cmd
Default_cmd = DefaultCmd.get_default_cmd()
find = 0
timeout = ''
pod_inject = ''
(target_host, is_exist) = get_target_host(dto)
if not is_exist:
return 'Host: ' + target_host + ' does not exist.'
j = random.randint(0, len(Default_cmd) - 1)
inject_type = Default_cmd.keys()[j]
if inject_type == "k8s":
pod_list = K8sObserver.get_pod_name_list("sock-shop")
k = random.randint(0, len(pod_list) - 1)
pod_inject = pod_list[k]
target_inject = Default_cmd[Default_cmd.keys()
[j]] + " " + pod_inject
else:
target_inject = Default_cmd[Default_cmd.keys()[j]]
if dto['timeout'] == 'default':
timeout = ' --timeout 300'
elif dto['timeout'] != 'no':
timeout = ' --timeout ' + dto['timeout']
for i in range(0, len(has_injected)):
if has_injected[i]['host'] == target_host and has_injected[i][
'inject_type'] == target_inject:
find = 1
if find == 0:
if inject_type != "k8s":
target_inject = target_inject + timeout
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=target_inject)
result = r.get_adhoc_result()
return handle_inject_result(inject_type, target_host,
target_inject, result,
sys._getframe().f_code.co_name,
dto['open'])
else:
Logger.log(
"error", "HOST HAS BEEN INJECTED - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(dto))
return "The host has been injected by the inject"
def get_pods_status(namespace):
r = Runner()
r.run_ad_hoc(hosts="10.60.38.181",
module='shell',
args='/opt/kube/bin/kubectl get pods -n ' + namespace)
result = r.get_adhoc_result()
result_status = []
if len(result['success']) > 0:
host_name = result['success'].keys()[0]
stdout_lines = result['success'][host_name]['stdout_lines']
for index in range(1, len(stdout_lines)):
stdout_line = stdout_lines[index].split()
result_status.append({
'name': stdout_line[0],
'status': stdout_line[2],
'restarts': stdout_line[3],
'age': stdout_line[4],
})
return result_status
def stop_all_on_specific_node(dto):
(target_host, is_exist) = get_target_host(dto)
if not is_exist:
return 'Host: ' + target_host + ' does not exist.'
target_inject = './blade status --type create'
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=target_inject)
result = r.get_adhoc_result()
uid_list = []
result_list = []
if len(result["success"]) > 0:
transform_ip = result["success"].keys()[0]
info = json.loads(result["success"][transform_ip]["stdout"].encode(
'unicode-escape').decode('string_escape'))["result"]
for i in info:
if i["Status"] == "Success":
uid_list.append(i["Uid"])
if len(uid_list) == 0:
Logger.log(
'info', 'NO CHAOS INJECT - Method : ' +
sys._getframe().f_code.co_name + "() - - " + str(dto))
return 'There is no injected attack on this host' + target_host
for item in uid_list:
cmd = './blade destroy ' + item
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=cmd)
result = r.get_adhoc_result()
# if len(result["success"]) > 0:
# for i in range(0, len(inject_info)):
# if inject_info[i]['cmd_id'] == item:
# if target_host == inject_info[i]['ip']:
# inject_info.pop(i)
# for i in range(0, len(has_injected)):
# if target_host == has_injected[i]['host'] and item == has_injected[i]['tag']:
# has_injected.pop(i)
# break
result_list.append(
handle_inject_result("destroy", target_host, cmd, result,
sys._getframe().f_code.co_name,
dto['open']))
return result_list
def chaos_inject_pod_single(dto):
find = 0
(target_host, is_exist) = get_target_host(dto)
if not is_exist:
return 'Host: ' + target_host + ' does not exist.'
target_inject = Cmd['k8s'] + dto['pod']
for i in range(0, len(has_injected)):
if has_injected[i]['host'] == target_host and has_injected[i][
'inject_type'] == target_inject:
find = 1
if find == 0:
r = Runner()
r.run_ad_hoc(hosts=target_host, module='shell', args=target_inject)
result = r.get_adhoc_result()
print(result)
return handle_inject_result('k8s', target_host, target_inject,
result,
sys._getframe().f_code.co_name,
dto['open'])
else:
return 'The pod has been injected'
def get_pod_name_list(namespace):
"""
获取某namespace下所有的pods名称
注意: 该函数只供fault_injector.py调用
:param namespace
:return: pods name list
"""
r = Runner()
r.run_ad_hoc(hosts="10.60.38.181",
module='shell',
args="/opt/kube/bin/kubectl get pods -n " + namespace)
result = r.get_adhoc_result()
name_list = []
if len(result["success"]) > 0:
transform_ip = result["success"].keys()[0]
stdout_lines = result["success"][transform_ip]["stdout_lines"]
for line in stdout_lines[1:]:
line = line.split()
name = line[0]
name_list.append(name)
return name_list
def stop_specific_chaos_inject(dto):
stop_id = dto['tag']
target_host = ''
find = 0
key = 0
for i in range(0, len(inject_info)):
if inject_info[i]['cmd_id'] == stop_id:
target_host = inject_info[i]['ip']
find = 1
key = i
for i in range(0, len(has_injected)):
if target_host == has_injected[i][
'host'] and stop_id == has_injected[i]['tag']:
has_injected.pop(i)
break
if find == 1:
r = Runner()
r.run_ad_hoc(hosts=Spare_hosts[inject_info[key]['ip']],
module='shell',
args='./blade destroy ' + stop_id)
result = r.get_adhoc_result()
if len(result["success"]) > 0:
transform_ip = result["success"].keys()[0]
the_stop_info = {
"position": inject_info[key]["position"],
"ip": Spare_hosts[inject_info[key]["ip"]],
"start_time": result["success"][transform_ip]["start"],
"cmd": result["success"][transform_ip]["cmd"],
}
inject_info.pop(key)
Logger.log(
'info',
'SUCCESS - Method : ' + sys._getframe().f_code.co_name +
"() - - " + str(the_stop_info))
if dto['open'] == 'true':
RabbitMq.connect(the_stop_info)
return result
else:
if len(result["unreachable"]) > 0:
transform_ip = result["unreachable"].keys()[0]
message = result["unreachable"][transform_ip]["msg"]
flag = "UNREACHABLE"
else:
transform_ip = result["failed"].keys()[0]
message = result["failed"][transform_ip]["msg"]
flag = "FAILED"
the_stop_info = {
"position": inject_info[key]["position"],
"ip": Spare_hosts[inject_info[key]["ip"]],
"cmd": "./blade destroy " + stop_id,
"message": message
}
Logger.log(
"error",
flag + " - Method : " + sys._getframe().f_code.co_name +
"() - - " + str(the_stop_info))
return result
else:
the_stop_info = {
"cmd": "./blade destroy " + stop_id,
}
Logger.log(
'error',
'UID NOT FOUND - Method : ' + sys._getframe().f_code.co_name +
"() - - " + str(the_stop_info))
return 'Inject not found'
def get_service_log(service, namespace):
"""
获取某个namespace下, 指定service的全部日志
:param service: service name
:param namespace: namespace
:return: success or failed
"""
if service not in service_name:
info = {
"error":
"The service does not exist. Please check the service name!",
}
Logger.log(
"error", "SERVICE NOT EXIST - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(info))
return "The service" + service + " does not exist. Please check the service name!"
r = Runner()
r.run_ad_hoc(hosts="10.60.38.181",
module='shell',
args="/opt/kube/bin/kubectl get pods -n " + namespace)
result = r.get_adhoc_result()
name_list = []
if len(result["success"]) > 0:
transform_ip = result["success"].keys()[0]
stdout_lines = result["success"][transform_ip]["stdout_lines"]
for line in stdout_lines[1:]:
if service in line and (service + "-db") not in line and (
service + "-external") not in line:
line = line.split()
name = line[0]
name_list.append(name)
else:
continue
print name_list
if len(name_list) != 0:
info = {
"success": "The " + service +
"service pod list was successfully obtained!",
"pod": name_list
}
Logger.log(
"info", "SUCCESS - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(info))
else:
info = {
"error": "The service" + service + " has no pod list!",
}
Logger.log(
"error", "NO POD - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(info))
return "The service" + service + " has no pod list!"
else:
if len(result["unreachable"]) > 0:
transform_ip = result["unreachable"].keys()[0]
message = result["unreachable"][transform_ip]["msg"]
flag = "UNREACHABLE"
else:
transform_ip = result["failed"].keys()[0]
message = result["failed"][transform_ip]["msg"]
flag = "FAILED"
info = {
"error": "Failed to get the pod list of service " + service,
"message": message
}
Logger.log(
"error", flag + " - Method : " +
sys._getframe().f_code.co_name + "() - - " + str(info))
return "Failed to get the pod list of services " + service
request_time = time.strftime('%Y-%m-%d/%H:%M:%S',
time.localtime(time.time()))
result = []
for pod_name in name_list:
result.append(get_pod_log(pod_name, namespace, request_time))
return result