[
https://issues.apache.org/jira/browse/AMBARI-13424?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alejandro Fernandez updated AMBARI-13424:
-----------------------------------------
Description:
In some environments, the customer still wants to proceed with RU & EU even if
PreChecks fail. E.g.,
NameNode HA doesn't actually exist if they have their own distribution of HDFS
Hosts be down at the time the upgrade is starting, but will come online during
the upgrade.
Today, they are unable to start RU/EU if any of the PreChecks fail.
Introduce a config in ambari.properties file called
"stack.upgrade.bypass.prechecks" whose default value is "false"
was:
If the user disables preUpgradeCheck from the Experimental settings page, and
then presses the "Perform Upgrade" button, they are immediately prompted with a
preupgradecheck error, rather than proress of upgrade of HDFS. On Ambari 2.1.1,
upgrade commenced and proceeded until the HDFS component itself encountered an
error during upgrade.
This blocks rolling upgrade with Isilon. NameNode HA is not relevant. If the
preupgradecheck is ignored with Isilon, upgrade should commence and then the
HDFS upgrade step will complete. (eg, 2.1.0, 2.1.1.)
Here is the text within the box that appears (remember, this is with
preUpgradeCheck disabled).
Error
500 status code received on POST method for API:
/api/v1/clusters/Hello/upgrades
Error message: An internal system exception occurred: Unable to perform
upgrade. Prerequisite checks failed
[{"type":{"name":"PreUpgradeCheck","ordinal":62},"propertiesMap":{"UpgradeChecks":{"check":"NameNode
High Availability must be
enabled","check_type":"SERVICE","cluster_name":"Hello","failed_detail":[],"failed_on":["HDFS"],"id":"SERVICES_NAMENODE_HA","reason":"NameNode
High Availability is not enabled. Verify that dfs.nameservices property is
present in
hdfs-site.xml.","repository_version":"2.3.2.0-2950","status":"FAIL"}}}]
> Allow disabling RU/EU PreChecks
> -------------------------------
>
> Key: AMBARI-13424
> URL: https://issues.apache.org/jira/browse/AMBARI-13424
> Project: Ambari
> Issue Type: Bug
> Components: ambari-web
> Affects Versions: 2.1.2
> Environment: Ambari 2.1.2, upgrading HDP 2.2.8 to 2.3.0.0.
> Single CentOS VM running Ambari Server and Ambari Agent 2.1.2, and HDP.
> Only hdfs, zookeeper and ambari metrics components are installed.
> The cluster has no NameNode HA enabled.
> Reporter: Robert Ketcherside
> Assignee: Alejandro Fernandez
>
> In some environments, the customer still wants to proceed with RU & EU even
> if PreChecks fail. E.g.,
> NameNode HA doesn't actually exist if they have their own distribution of HDFS
> Hosts be down at the time the upgrade is starting, but will come online
> during the upgrade.
> Today, they are unable to start RU/EU if any of the PreChecks fail.
> Introduce a config in ambari.properties file called
> "stack.upgrade.bypass.prechecks" whose default value is "false"
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
