As i understood, most of the imported tables in the schema are empty, it
basically a reason why Ambari not starting.
For what you should do, verify the backup data - does it contain more
information than was imported (i.e. it is some import glitch) or the DB
was not fully exported.
If DB is totally lost, could be deployed a test mirror cluster (a bit
complex way and not guarantee anything):
- create a full mapping of services and components you have on each host
- create a test cluster with (ideally if you have blueprint for the
broken cluster):
- same host names,
- services & components location per host
- services user name, if custom were used
- cluster name
....(add here anything what was customized on the cluster)...
- same versions of the stack and Ambari
- apply same cluster configuration, like Kerberos, HDFS HA and so on
(complexity depends on amount of services and their custom
configuration/features used)
- make a DB snapshot of deployed and configured test cluster and apply
it to the production cluster with fixing after the host table with
proper ip addresses
- fix any pop-up issue after
It is quite complex way, but in theory should not mess with the data on
the cluster
------ Original Message ------
From: "Matt Andruff" <matt.andr...@gmail.com>
To: user@ambari.apache.org
Sent: 6/17/2020 3:41:43 PM
Subject: Re: reinitialize ambari
You can reset Ambari and delete the contents of the database with
./ambari-server reset
But it will flatten your installation and effectively be like starting
from the beginning, nothing is remembered. It's very destructive.
On Wed., Jun. 17, 2020, 03:10 Leoš Literák, <l...@literak.cz> wrote:
Hi,
my colleague has installed the Ambari + Hadoop stack on our servers.
He used Postgresql 11 to store the information. Then I had to
downgrade the Postgresql to 9.3.9. I have exported the ambari schema
from 11 to 9.3.9. Since then Ambari stopped working even when we
started Postgresql 11 again.
The current situation is that I can log in to Ambari and it shows an
empty installation with Launch wizard. Hadoop is stopped and there are
no manual scripts. The colleague is not available and I need to
continue.
It is Ambari 2.6.2.2.
ambari-server.out
Database consistency check result: DB_CHECK_WARNING
ambari-server.log
17 Jun 2020 07:10:43,169 INFO [main] ControllerModule:224 - Detected
POSTGRES as the database type from the JDBC URL
17 Jun 2020 07:10:43,181 INFO [main] ControllerModule:267 - Using
c3p0 ComboPooledDataSource as the EclipsLink DataSource
17 Jun 2020 07:10:52,612 INFO [main] StackModule:191 - Resolve:
HDP:2.6
17 Jun 2020 07:10:52,612 INFO [main] StackModule:374 -
mergeServiceWithExplicitParentcommon-services/DRUID/0.10.1
17 Jun 2020 07:10:52,612 INFO [main] ServiceModule:185 - Resolve
service
17 Jun 2020 07:10:52,612 INFO [main] ServiceModule:198 - Display name
service/parent: null/Druid
17 Jun 2020 07:11:06,165 INFO [main] WebApplicationImpl:815 -
Initiating Jersey application, version 'Jersey: 1.19 02/11/2015 03:25
AM'
17 Jun 2020 07:11:06,310 WARN [main] Errors:173 - The following
warnings have been detected with resource and/or provider classes:
WARNING: A HTTP GET method, public javax.ws.rs.core.Response
org.apache.ambari.server.api.services.ActionService.getActionDefinitions(java.lang.String,javax.ws.rs.core.HttpHeaders,javax.ws.rs.core.UriInfo),
should not consume any entity.
17 Jun 2020 07:11:06,427 INFO [main] AmbariServer:565 - *********
Started Server **********
17 Jun 2020 07:12:15,345 ERROR [alert-event-bus-1]
AlertReceivedListener:480 - Unable to process alert
yarn_resourcemanager_webui for an invalid cluster named
centaur_sandbox
17 Jun 2020 07:12:15,345 ERROR [alert-event-bus-2]
AlertReceivedListener:480 - Unable to process alert
ams_metrics_collector_process for an invalid cluster named
centaur_sandbox
17 Jun 2020 08:46:06,661 INFO [pool-18-thread-1]
MetricsServiceImpl:65 - Attempting to initialize metrics sink
17 Jun 2020 08:46:06,661 INFO [pool-18-thread-1]
MetricsServiceImpl:81 - ********* Configuring Metric Sink **********
17 Jun 2020 08:46:06,661 INFO [pool-18-thread-1]
AmbariMetricSinkImpl:95 - No clusters configured.
17 Jun 2020 08:46:13,343 ERROR [alert-event-bus-2]
AlertReceivedListener:480 - Unable to process alert
ams_metrics_collector_process for an invalid cluster named
centaur_sandbox
17 Jun 2020 08:46:13,343 ERROR [alert-event-bus-2]
AlertReceivedListener:480 - Unable to process alert
ambari_agent_disk_usage for an invalid cluster named centaur_sandbox
17 Jun 2020 08:46:13,343 ERROR [alert-event-bus-2]
AlertReceivedListener:480 - Unable to process alert datanode_webui for
an invalid cluster named centaur_sandbox
17 Jun 2020 08:46:13,344 ERROR [alert-event-bus-2]
AlertReceivedListener:480 - Unable to process alert
ams_metrics_collector_autostart for an invalid cluster named
centaur_sandbox
select * from host returns 3 rows
ambari=> select * from stack;
stack_id | stack_name | stack_version
----------+-------------+-----------------
1 | HDP | 2.0.6
2 | HDP | 2.3.ECS
16 | BigInsights | 4.2
(16 rows)
Tables groups, clusters are empty. Well, most of the tables are empty.
If the ambari schema is broken, is there any reasonable way to
populate it again from the existing cluster? Or does it mean we have
to remove all components and start the wizard from scratch?
Thank you
Leos
