web-dev-qa-db-ja.com

ランドスケープインストールを伴うOpenstackが失敗する

このサイトの指示に従いました: http://www.openstackbasement.com/

「Sudo openstack-install」を発行するまで、最後まで到達しますが、かなり遠くに行きますが、次のページに残っています: openstack-fail

約2900秒後に失敗します。以下は、MAASのコントローラーのエラーログです。 http://Pastebin.com/raw/A7qtJm4v (1)

展開されたノードのエラーログは次のようになります。 http://Pastebin.com/raw/zuX1TJcB (2)

デプロイされたノードの画面は次のようになります。 openstack-fail2

質問:なぜこれが引き起こされ、どうすれば解決できますか?

追伸ポイント2の最初のテキストをスキップする必要がありました。それ以外の場合は、投稿するのに時間がかかりました。 Pastebinにはファイル全体が含まれています。

1)

[INFO: 09-05 23:32:16, openstack-install:227] Starting OpenStack Installer v0.99.28
[INFO: 09-05 23:32:16, openstack-install:228] Start command: ['/usr/bin/openstack-install']
[INFO: 09-05 23:32:16, openstack-install:239] Creating juju directories: /home/[user]/.cloud-install/juju
[INFO: 09-05 23:32:19, openstack-install:295] Running Liberty release
[INFO: 09-05 23:32:32, installbase.py:132] Performing an Autopilot install
[INFO: 09-05 23:32:32, utils.py:780] pollinate: Sudo su - -c 'pollinate -q -r --curl-opts "-k --user-agent uoi/bb2f64a2-82d8-4196-b55b-1f42c0141c1b/IL"'
[DEBUG: 09-05 23:32:58, landscape.py:75] Existing MAAS defined, doing a LDS installation with existing MAAS.
[DEBUG: 09-05 23:33:14, utils.py:627] ssh keys exist for this user, they will be used instead.
[DEBUG: 09-05 23:33:14, multi.py:139] Bootstrapping Juju: JUJU_HOME=/home/[user]/.cloud-install/juju juju  bootstrap 
[ERROR: 09-05 23:38:49, multi.py:218] Failed to get ip directly: [Errno -2] Name or service not known
[DEBUG: 09-05 23:38:51, multi.py:177] Finished MAAS step, now deploying Landscape.
[INFO: 09-06 00:27:05, utils.py:780] pollinate: Sudo su - -c 'pollinate -q -r --curl-opts "-k --user-agent uoi/bb2f64a2-82d8-4196-b55b-1f42c0141c1b/ET"'
[ERROR: 09-06 00:27:05, multi.py:384] Problem deploying Landscape: {'err': "2016-09-05 23:39:04 
[DEBUG] deployer.cli: Using runtime GoEnvironment on maas\n2016-09-05 23:39:04 
[INFO] deployer.cli: Starting deployment of landscape-dense-maas\n2016-09-05 23:39:04 
[DEBUG] deployer.import: Getting charms...\n2016-09-05 23:39:04 [DEBUG] deployer.charm: Cache dir /home/[user]/.cloud-install/juju/.deployer-store-cache/cs_trusty_haproxy-16\n2016-09-05 23:39:04 
[DEBUG] deployer.charm: Retrieving store charm cs:trusty/haproxy-16\n2016-09-05 23:39:04 [DEBUG] deployer.charm: Cache dir /home/[user]/.cloud-install/juju/.deployer-store-cache/cs_trusty_landscape-server\n2016-09-05 23:39:04 
[DEBUG] deployer.charm: Retrieving store charm cs:trusty/landscape-server-15\n2016-09-05 23:39:04 
[DEBUG] deployer.charm: Cache dir /home/[user]/.cloud-install/juju/.deployer-store-cache/cs_trusty_postgresql-40\n2016-09-05 23:39:04       
[DEBUG] deployer.charm: Retrieving store charm cs:trusty/postgresql-40\n2016-09-05 23:39:04 
[DEBUG] deployer.charm: Cache dir /home/[user]/.cloud-install/juju/.deployer-store-cache/cs_trusty_rabbitmq-server-43\n2016-09-05 23:39:04     
[DEBUG] deployer.charm: Retrieving store charm cs:trusty/rabbitmq-server-43\n2016-09-05 23:39:05 
[DEBUG] deployer.deploy: Resolving configuration\n2016-09-05 23:39:05      
[DEBUG] deployer.env: Connecting to environment...\n2016-09-05 23:39:05    
[DEBUG] deployer.env: Connected to environment\n2016-09-05 23:39:05     
[INFO] deployer.import: Deploying services...\n2016-09-05 23:39:05         
[INFO] deployer.import:  Deploying service haproxy using cs:trusty/haproxy-16\n2016-09-05 23:39:05 [DEBUG] deployer.import:  Refetching status for placement deploys\n2016-09-05 23:39:14   
[DEBUG] deployer.import:  Setting annotations\n2016-09-05 23:39:15        
[INFO] deployer.import:  Deploying service landscape-server using cs:trusty/landscape-server\n2016-09-05 23:39:19 
[DEBUG] deployer.import:  Setting annotations\n2016-09-05 23:39:20 
[INFO] deployer.import:  Deploying service postgresql using cs:trusty/postgresql-40\n2016-09-05 23:39:25 
[DEBUG] deployer.import:  Setting annotations\n2016-09-05 23:39:25       
[INFO] deployer.import:  Deploying service rabbitmq-server using cs:trusty/rabbitmq-server-43\n2016-09-05 23:39:29 
[DEBUG] deployer.import:  Setting annotations\n2016-09-05 23:39:34  
[DEBUG] deployer.import: Adding units...\n2016-09-05 23:39:34 
[DEBUG] deployer.import:  Service 'haproxy' does not need any more units added.\n2016-09-05 23:39:34 [DEBUG] deployer.import:  Service 'landscape-server' does not need any more units added.\n2016-09-05 23:39:34 
[DEBUG] deployer.import:  Service 'postgresql' does not need any more units added.\n2016-09-05 23:39:34     
[DEBUG] deployer.import:  Service 'rabbitmq-server' does not need any more units added.\n2016-09-05 23:39:34 
[DEBUG] deployer.import: Waiting for units before adding relations\n2016-09-05 23:46:02     
[DEBUG] deployer.env:  Delta machine: 0/lxc/0 change:pending\n2016-09-05 23:47:03 [DEBUG] deployer.env:  Delta machine: 0/lxc/1 change:pending\n2016-09-05 23:48:03     
[DEBUG] deployer.env:  Delta machine: 0/lxc/2 change:pending\n2016-09-05 23:49:11 [DEBUG] deployer.env:  Delta machine: 0/lxc/3 change:pending\n2016-09-06 00:24:03 
[DEBUG] deployer.env: Connecting to environment...\n2016-09-06 00:24:04    
[DEBUG] deployer.env: Connected to environment\n2016-09-06 00:24:04  
[INFO] deployer.import: Adding relations...\n2016-09-06 00:24:04 
[INFO] deployer.import:  Adding relation landscape-server <-> rabbitmq-server\n2016-09-06 00:24:04 
[INFO] deployer.import:  Adding relation landscape-server <-> haproxy\n2016-09-06 00:24:04 
[INFO] deployer.import:  Adding relation landscape-server:db <-> postgresql:db-admin\n2016-09-06 00:24:05 
[DEBUG] deployer.import: Waiting for relation convergence 180s\n2016-09-06 00:27:05 
[ERROR] deployer.import: Reached deployment timeout.. exiting\n2016-09-06 00:27:05     
[INFO] deployer.cli: Deployment stopped. run time: 2881.35\n", 'output': '', 'status': 1}
[ERROR: 09-06 00:27:05, gui.py:269] A fatal error has occurred: Error deploying Landscape.

[ERROR: 09-06 00:27:05, gui.py:270] Error deploying Landscape.
Traceback (most recent call last):
File "/usr/lib/python3.4/concurrent/futures/thread.py", line 54, in run
result = self.fn(*self.args, **self.kwargs)
File "/usr/share/openstack/cloudinstall/controllers/install/multi.py", line 181, in do_install
self.loop).run()
File "/usr/share/openstack/cloudinstall/controllers/install/multi.py", line 319, in run
self.deploy_landscape()
File "/usr/share/openstack/cloudinstall/controllers/install/multi.py", line 344, in deploy_landscape
self.run_deployer()
File "/usr/share/openstack/cloudinstall/controllers/install/multi.py", line 385, in run_deployer
raise Exception("Error deploying Landscape.")
Exception: Error deploying Landscape.
[DEBUG: 09-06 00:27:05, error.py:35] showing error view for: Error deploying Landscape.

2)

2016-09-06 08:01:45 INFO juju.mongo open.go:125 dialled mongo successfully on address "127.0.0.1:37017"
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":20,"Type":"Rsyslog","Request":"GetRsyslogConfig","Params":"'params redacted'"}
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":21,"Type":"Provisioner","Version":1,"Request":"Life","Params":"'params redacted'"}
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 146.525279ms {"RequestId":13,"Response":"'body redacted'"} Machiner[""].Life
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 147.19989ms {"RequestId":14,"Response":"'body redacted'"} Reboot[""].WatchForRebootEvent
2016-09-06 08:01:45 INFO juju.mongo open.go:125 dialled mongo successfully on address "192.168.1.151:37017"
2016-09-06 08:01:45 INFO juju.mongo open.go:125 dialled mongo successfully on address "192.168.1.151:37017"
2016-09-06 08:01:45 INFO juju.mongo open.go:125 dialled mongo successfully on address "192.168.1.151:37017"
2016-09-06 08:01:45 INFO juju.mongo open.go:125 dialled mongo successfully on address "192.168.1.151:37017"
2016-09-06 08:01:45 DEBUG juju.network network.go:268 no lxc bridge addresses to filter for machine
2016-09-06 08:01:45 INFO juju.worker.machiner machiner.go:132 setting addresses for machine-0 to ["local-machine:127.0.0.1" "local-cloud:192.168.1.151" "local-machine:::1"]
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 136.703326ms {"RequestId":15,"Response":"'body redacted'"} Machiner[""].WatchAPIHostPorts
2016-09-06 08:01:45 INFO juju.mongo open.go:125 dialled mongo successfully on address "192.168.1.151:37017"
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":22,"Type":"Machiner","Request":"SetMachineAddresses","Params":"'params redacted'"}
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":23,"Type":"Reboot","Version":1,"Request":"GetRebootAction","Params":"'params redacted'"}
2016-09-06 08:01:45 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":24,"Type":"NotifyWatcher","Id":"3","Request":"Next","Params":"'params redacted'"}
2016-09-06 08:01:45 DEBUG juju.state open.go:57 connection established
2016-09-06 08:01:45 DEBUG juju.worker.peergrouper worker.go:432 found new machine "0"
2016-09-06 08:01:45 DEBUG juju.state open.go:64 mongodb login successful
2016-09-06 08:01:45 INFO juju.worker.diskmanager diskmanager.go:62 block devices changed: [{sda [/dev/disk/by-path/pci-0000:00:10.0-scsi-0:0:0:0]    scsi@32:0.0.0 25600  true } {sdb [/dev/disk/by-path/pci-0000:00:10.0-scsi-0:0:1:0]    scsi@32:0.1.0 25600  false }]
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 230.408525ms {"RequestId":16,"Response":"'body redacted'"} ProxyUpdater[""].WatchForProxyConfigAndAPIHostPortChanges
2016-09-06 08:01:46 INFO juju.mongo open.go:125 dialled mongo successfully on address "127.0.0.1:37017"
2016-09-06 08:01:46 INFO juju.mongo open.go:125 dialled mongo successfully on address "127.0.0.1:37017"
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":25,"Type":"Machiner","Request":"APIHostPorts","Params":"'params redacted'"}
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":26,"Type":"NotifyWatcher","Id":"4","Request":"Next","Params":"'params redacted'"}
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":27,"Type":"DiskManager","Version":1,"Request":"SetMachineBlockDevices","Params":"'params redacted'"}
2016-09-06 08:01:46 INFO juju.mongo open.go:125 dialled mongo successfully on address "127.0.0.1:37017"
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 216.92964ms {"RequestId":21,"Response":"'body redacted'"} Provisioner[""].Life
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":28,"Type":"ProxyUpdater","Version":1,"Request":"ProxyConfig","Params":"'params redacted'"}
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":29,"Type":"NotifyWatcher","Id":"5","Request":"Next","Params":"'params redacted'"}
2016-09-06 08:01:46 INFO juju.cmd.jujud machine.go:1092 update apiserver worker with new certificate
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 37.154757ms {"RequestId":25,"Response":"'body redacted'"} Machiner[""].APIHostPorts
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 139.568869ms {"RequestId":23,"Response":"'body redacted'"} Reboot[""].GetRebootAction
2016-09-06 08:01:46 INFO juju.worker.certupdater certupdater.go:175 State Server cerificate addresses updated to ["192.168.1.151" "anything" "juju-apiserver" "juju-mongodb" "localhost"]
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:262 <- [4] machine-0 {"RequestId":30,"Type":"Provisioner","Version":1,"Request":"SetSupportedContainers","Params":"'params redacted'"}
2016-09-06 08:01:46 INFO juju.apiserver apiserver.go:143 updating api server certificate
2016-09-06 08:01:46 INFO juju.apiserver apiserver.go:150 new certificate addresses: 192.168.1.151
2016-09-06 08:01:46 DEBUG juju.network network.go:268 no lxc bridge addresses to filter for machine
2016-09-06 08:01:46 INFO juju.agent agent.go:565 API server address details [["node2.maas:17070" "192.168.1.151:17070"]] written to agent config as ["192.168.1.151:17070"]
2016-09-06 08:01:46 DEBUG juju.worker.reboot reboot.go:67 Reboot worker got action: noop
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 323.212726ms {"RequestId":17,"Response":"'body redacted'"} Logger[""].LoggingConfig
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 322.960531ms {"RequestId":19,"Response":"'body redacted'"} KeyUpdater[""].AuthorisedKeys
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 323.659437ms {"RequestId":18,"Response":"'body redacted'"} StorageProvisioner[""].WatchForEnvironConfigChanges
2016-09-06 08:01:46 DEBUG juju.apiserver apiserver.go:276 -> [4] machine-0 52.597332ms {"RequestId":28,"Response":"'body redacted'"} ProxyUpdater[""].ProxyConfig
2016-09-06 08:01:46 DEBUG juju.worker.logger logger.go:45 reconfiguring logging from "<root>=DEBUG" to "<root>=WARNING;unit=DEBUG"
2016-09-06 08:02:17 ERROR juju.state.unit unit.go:738 unit haproxy/0 cannot get assigned machine: unit "haproxy/0" is not assigned to a machine
2016-09-06 08:02:17 WARNING juju.state allwatcher.go:351 getting a public  address for unit "haproxy/0" failed: "unit haproxy/0 cannot get assigned machine: unit \"haproxy/0\" is not assigned to a machine"
2016-09-06 08:02:17 ERROR juju.state.unit unit.go:748 unit haproxy/0 cannot get assigned machine: unit "haproxy/0" is not assigned to a machine
2016-09-06 08:02:17 WARNING juju.state allwatcher.go:355 getting a private address for unit "haproxy/0" failed: "unit haproxy/0 cannot get assigned machine: unit \"haproxy/0\" is not assigned to a machine"
2016-09-06 08:02:22 WARNING juju.state allwatcher.go:351 getting a public address for unit "haproxy/0" failed: "public no address"
2016-09-06 08:02:22 WARNING juju.state allwatcher.go:355 getting a private address for unit "haproxy/0" failed: "private no address"
2016-09-06 08:02:22 ERROR juju.state.unit unit.go:738 unit landscape-server/0 cannot get assigned machine: unit "landscape-server/0" is not assigned to a machine
2016-09-06 08:02:22 WARNING juju.state allwatcher.go:351 getting a public address for unit "landscape-server/0" failed: "unit landscape-server/0 cannot get assigned machine: unit \"landscape-server/0\" is not assigned to a machine"
2016-09-06 08:02:22 ERROR juju.state.unit unit.go:748 unit landscape-server/0 cannot get assigned machine: unit "landscape-server/0" is not assigned to a machine
2016-09-06 08:02:22 WARNING juju.state allwatcher.go:355 getting a private address for unit "landscape-server/0" failed: "unit landscape-server/0 cannot get assigned machine: unit \"landscape-server/0\" is not assigned to a machine"
2016-09-06 08:02:27 WARNING juju.state allwatcher.go:351 getting a public address for unit "landscape-server/0" failed: "public no address"
2016-09-06 08:02:27 WARNING juju.state allwatcher.go:355 getting a private address for unit "landscape-server/0" failed: "private no address"
2016-09-06 08:02:27 ERROR juju.state.unit unit.go:738 unit postgresql/0 cannot get assigned machine: unit "postgresql/0" is not assigned to a machine
2016-09-06 08:02:27 WARNING juju.state allwatcher.go:351 getting a public address for unit "postgresql/0" failed: "unit postgresql/0 cannot get assigned machine: unit \"postgresql/0\" is not assigned to a machine"
2016-09-06 08:02:27 ERROR juju.state.unit unit.go:748 unit postgresql/0 cannot get assigned machine: unit "postgresql/0" is not assigned to a machine
2016-09-06 08:02:27 WARNING juju.state allwatcher.go:355 getting a private address for unit "postgresql/0" failed: "unit postgresql/0 cannot get assigned machine: unit \"postgresql/0\" is not assigned to a machine"
2016-09-06 08:02:32 WARNING juju.state allwatcher.go:351 getting a public address for unit "postgresql/0" failed: "public no address"
2016-09-06 08:02:32 WARNING juju.state allwatcher.go:355 getting a private address for unit "postgresql/0" failed: "private no address"
2016-09-06 08:02:32 ERROR juju.state.unit unit.go:738 unit rabbitmq-server/0 cannot get assigned machine: unit "rabbitmq-server/0" is not assigned to a machine
2016-09-06 08:02:32 WARNING juju.state allwatcher.go:351 getting a public address for unit "rabbitmq-server/0" failed: "unit rabbitmq-server/0 cannot get assigned machine: unit \"rabbitmq-server/0\" is not assigned to a machine"
2016-09-06 08:02:32 ERROR juju.state.unit unit.go:748 unit rabbitmq-server/0 cannot get assigned machine: unit "rabbitmq-server/0" is not assigned to a machine
2016-09-06 08:02:32 WARNING juju.state allwatcher.go:355 getting a private address for unit "rabbitmq-server/0" failed: "unit rabbitmq-server/0 cannot get assigned machine: unit \"rabbitmq-server/0\" is not assigned to a machine"
2016-09-06 08:02:37 WARNING juju.state allwatcher.go:351 getting a public address for unit "rabbitmq-server/0" failed: "public no address"
2016-09-06 08:02:37 WARNING juju.state allwatcher.go:355 getting a private address for unit "rabbitmq-server/0" failed: "private no address"
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:465 error fetching public address: "public no address"
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:465 error fetching public address: "public no address"
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:465 error fetching public address: "public no address"
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:465 error fetching public address: "public no address"
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:679 error fetching public address: public no address
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:679 error fetching public address: public no address
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:679 error fetching public address: public no address
2016-09-06 16:16:56 WARNING juju.apiserver.client status.go:679 error fetching public address: public no address
4
user3892683

私はついに問題の答えを得ました。単一のVMware ESXIサーバーを使用して、ノードとコントローラーの両方を実行していました。いくつかのトラブルシューティングの後、JUJUがLXCでコンテナを起動していましたが、コンテナはIPアドレスを取得していませんでした。その時点で、問題の原因がJUJUかLXCかは不明でした。

インターネットでいくつか検索したところ、次のリンクを見つけた後、どちらも問題ではないことが判明しました: https://www.reddit.com/r/homelab/comments/4p3k9j/trouble_getting_lxc_networking_up_containers_not/

Ubuntu VM ESXi内でlxcを使用してしばらく前に同様の問題が発生しました。私の問題は、ESXiの仮想スイッチが無差別モードではなかったためです。落とされます。

VMware ESXIでプロミスキャスモードをオンにした後、コンテナはそこにIPアドレスを取得し、openstackインストーラーは成功しました。

0
user3892683

これは、必ずしもopenstack-installまたはAutopilotの問題を示しているわけではありません。あなたがする必要があるのは、juju bootstrap; juju deploy ubuntuそして、それが現在のmaasセットアップで機能することを確認します。信頼性の低いことが知られているbmcとしてWake on Lanも使用していると言われました。

1
battlemidget