AnsibleUndefinedVariable: 'dict object' has no attribute
kuldeepkulkarni09 opened this issue · 9 comments
Basically running playbook from CM host and getting below error, tried lot of things however could not figure out what is going wrong, if it’s problem with variables in group_var or am I missing something.
TASK [scm : file] **********************************************************************************************************************************************************************************************************************************************************************************************************************************************************
changed: [kkulkani-cdhkerberos-1]
TASK [scm : Import KDC admin credentials] **********************************************************************************************************************************************************************************************************************************************************************************************************************************
ok: [kkulkani-cdhkerberos-1]
TASK [scm : Wait for agent heartbeats] *************************************************************************************************************************************************************************************************************************************************************************************************************************************
Pausing for 30 seconds
(ctrl+C then 'C' = continue early, ctrl+C then 'A' = abort)
Press 'C' to continue the play or 'A' to abort
ok: [kkulkani-cdhkerberos-1]
TASK [scm : Prepare CMS template] ******************************************************************************************************************************************************************************************************************************************************************************************************************************************
fatal: [kkulkani-cdhkerberos-1]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'dict object' has no attribute u'kkulkani-cdhkerberos-1'"}
Same problem here!
ok: [xxx.xxx.xxx.xxx]
TASK [scm : Wait for agent heartbeats] ******************************************************************************************************************************
Pausing for 30 seconds
(ctrl+C then 'C' = continue early, ctrl+C then 'A' = abort)
ok: [xxx.xxx.xxx.xxx]
TASK [scm : Prepare CMS template] ***********************************************************************************************************************************
fatal: [xxx.xxx.xxx.xxx -> localhost]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'dict object' has no attribute u'xxx.xxx.xxx.xxx'"}
to retry, use: --limit @/opt/ansible/cloudera-playbook/site.retry
PLAY RECAP **********************************************************************************************************************************************************
xxx.xxx.xxx.xxx : ok=21 changed=2 unreachable=0 failed=1
Hi @rouxero, @kuldeepkulkarni09,
I have just tested the latest git repo. It works for me:
TASK [scm : Wait for agent heartbeats] *********************************************************************************
Pausing for 30 seconds
(ctrl+C then 'C' = continue early, ctrl+C then 'A' = abort)
ok: [playbook-test-1.gce.cloudera.com]
TASK [scm : Prepare CMS template] **************************************************************************************
changed: [playbook-test-1.gce.cloudera.com -> localhost]
TASK [scm : Setup the Cloudera Management Services (CMS)] **************************************************************
ok: [playbook-test-1.gce.cloudera.com -> localhost]
TASK [scm : debug] *****************************************************************************************************
ok: [playbook-test-1.gce.cloudera.com] => {
Could you please share with me your ansible_hosts configuration? Maybe something is different in your case.
encounter same issue and used the latest git repo.
I set krb5 type to none, ad and mit - same output.
Debug output
TASK [scm : Prepare CMS template] ******************************************************************************************************************************************************************** task path: /home/admin/cloudera-playbook/roles/scm/tasks/cms.yml:9 <localhost> ESTABLISH LOCAL CONNECTION FOR USER: admin <localhost> EXEC /bin/sh -c 'echo ~admin && sleep 0' <localhost> EXEC /bin/sh -c '( umask 77 && mkdir -p " echo /home/admin/.ansible/tmp/ansible-tmp-1557208095.46-187677447756307 " && echo ansible-tmp-1557208095.46-187677447756307=" echo /home/admin/.ansible/tmp/ansible-tmp-1557208095.46-187677447756307 " ) && sleep 0' <localhost> EXEC /bin/sh -c 'rm -f -r /home/admin/.ansible/tmp/ansible-tmp-1557208095.46-187677447756307/ > /dev/null 2>&1 && sleep 0' fatal: [server-admin-01 -> localhost]: FAILED! => { "changed": false, "msg": "AnsibleUndefinedVariable: 'dict object' has no attribute u'server-admin-01'" } to retry, use: --limit @/home/admin/cloudera-playbook/site.retry
Hostfile
`# Note for AWS: 'Public DNS' name is too long for ansible_host, use 'Public IP' (ansible/ansible#11536)
[scm_server]
server-admin-01 license_file=/home/admin/cloudera_license.txt
[db_server]
server-admin-01
[utility_servers:children]
scm_server
db_server
[krb5_server]
server-admin-01 default_realm=CLOUDERA.COM
[master_servers]
server-admin-02 host_template=HostTemplate-Master1
server-data-01 host_template=HostTemplate-Master2
server-data-01 host_template=HostTemplate-Master3
[worker_servers]
server-data-01
server-data-02
server-admin-02
[worker_servers:vars]
host_template=HostTemplate-Workers
[cdh_servers:children]
utility_servers
master_servers
krb5_server
worker_servers
[all:vars]
ansible_user=admin`
Hi @roczei ! any update on this issue. I am also getting the same. It seems to come after adding this --
[worker_servers:vars]
host_template=HostTemplate-Workers
Hi All,
I was able to reproduce this issue with alcher's ansible_hosts configuration. The root cause is that hostnames are not fully qualified domain names (FQDN). Here is an example for a FQDN: playbook-1.gce.cloudera.com
I have modified them to FQDNs (example: playbook-1 ---> playbook-1.gce.cloudera.com).
This was the next error because the HDFS-HTTPFS-1 was not there:
> TASK [cdh : Import cluster template] *********************************************************************************************
> fatal: [playbook-1.gce.cloudera.com]: FAILED! => {"changed": false, "connection": "close", "content": "{\n \"message\" : \"The following roles are not resolved by any host: HDFS-HTTPFS-1.\"\n}", "content_type": "application/json", "date": "Tue, 28 May 2019 21:06:46 GMT", "elapsed": 2, "expires": "Thu, 01-Jan-1970 00:00:00 GMT", "json": {"message": "The following roles are not resolved by any host: HDFS-HTTPFS-1."}, "msg": "Status code was 400 and not [200]: HTTP Error 400: Bad Request", "redirected": false, "server": "Jetty(6.1.26.cloudera.4)", "set_cookie": "CLOUDERA_MANAGER_SESSIONID=x4l8jw2qr0edxb8nb3k28d3g;Path=/;HttpOnly", "status": 400, "url": "http://playbook-1.gce.cloudera.com:7180/api/v19/cm/importClusterTemplate?addRepositories=true"}
Here is a working example, please test this (you need just modify the hostnames and use FQDNs everywhere):
[scm_server]
playbook-test-1.gce.cloudera.com license_file=/root/cloudera_license.txt
[db_server]
playbook-test-1.gce.cloudera.com
[krb5_server]
playbook-test-1.gce.cloudera.com default_realm=EXAMPLE.COM
[utility_servers:children]
scm_server
db_server
krb5_server
[gateway_servers]
playbook-test-1.gce.cloudera.com host_template=HostTemplate-Gateway role_ref_names=HDFS-HTTPFS-1
[master_servers]
playbook-test-2.gce.cloudera.com host_template=HostTemplate-Master1
playbook-test-3.gce.cloudera.com host_template=HostTemplate-Master2
playbook-test-4.gce.cloudera.com host_template=HostTemplate-Master3
[worker_servers]
playbook-test-5.gce.cloudera.com
playbook-test-6.gce.cloudera.com
playbook-test-7.gce.cloudera.com
[worker_servers:vars]
host_template=HostTemplate-Workers
[cdh_servers:children]
utility_servers
gateway_servers
master_servers
worker_servers
Confirmed working when using fqdn
I have just updated the README.md file in the git repository and highlighted that fully qualified domain name (FQDN) is mandatory in the ansible_hosts file.
Hello. This issue would occur if the hostnames in inventory file are different to the hostnames reported by each node's Cloudera Manager agent. Using short names instead of FQDNs is one possibility, and DNS configuration (or misconfiguration) is another.
Closing for now -- if you have further issues, please re-open and provide debug logs of your playbook run, specifically the scm role (and use -vv verbosity settings)