Bug 901192 (JBPAPP6-1638) - CLONE - The command :reload does not check the new configuration, and will cause server crashes if that config is broken
Summary: CLONE - The command :reload does not check the new configuration, and will ca...
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: JBPAPP6-1638
Product: JBoss Enterprise Application Platform 6
Classification: JBoss
Component: Server
Version: 6.0.0
Hardware: Unspecified
OS: Unspecified
high
high
Target Milestone: ER2
: EAP 6.1.0
Assignee: Brian Stansberry
QA Contact:
URL: http://jira.jboss.org/jira/browse/JBP...
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2012-11-19 13:31 UTC by Tom Fonteyne
Modified: 2013-07-23 18:46 UTC (History)
7 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2012-11-20 13:47:11 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Red Hat Issue Tracker AS7-5976 0 Critical Resolved The command :reload does not check the new configuration, and will cause server crashes if that config is broken 2015-09-18 00:39:11 UTC
Red Hat Issue Tracker JBPAPP6-1638 0 Major Closed CLONE - The command :reload does not check the new configuration, and will cause server crashes if that config is broken 2015-09-18 00:39:11 UTC

Description Tom Fonteyne 2012-11-19 13:31:14 UTC
Help Desk Ticket Reference: https://c.na7.visual.force.com/apex/Case_View?id=500A000000BeaCj&sfdc.override=1
Steps to Reproduce: Have a host controller (can be the DC itself) with one server up and running.

Modify manually the host.xml, introducing a typo, for example remove a "<" from tag.

From the CLI, execute:   /host=master:reload

The result is the host controller AND the server process quitting.


It would seem reasonable that the "reload" at least checks the new xml for schema errors and rejects it.
project_key: JBPAPP6

The command :reload does not check the new configuration, and will cause server crashes if that config is broken

Comment 1 Tom Fonteyne 2012-11-19 13:31:15 UTC
Link: Added: This issue Cloned from AS7-5976


Comment 2 Tom Fonteyne 2012-11-19 13:35:02 UTC
Workflow: Removed: GIT Pull Request workflow  Added: jira
Security: Added: Public


Comment 4 Brian Stansberry 2012-11-19 15:05:07 UTC
First, editing the xml file underneath a running server is a bad practice we shouldn't attempt to support.

Second, it does check the xml – when it parses it. The only reason to do basic validation before shutting down the old services is to support people trying to manually edit the xml under a running server. For which, see above.

Comment 5 Tom Fonteyne 2012-12-10 09:16:25 UTC
reopening as per Brian's request

Comment 6 Rostislav Svoboda 2013-02-21 10:55:11 UTC
Is this item targeted to EAP 6.1.0 ? If yes, please add 'jboss-eap-6.1.0' flag.

Comment 7 Brian Stansberry 2013-02-21 21:02:01 UTC
This will be in EAP 6.1.0. It was fixed in upstream quite a while ago.

Comment 8 Jakub Cechacek 2013-03-29 13:24:07 UTC
Unless I'm doing something wrong the behavior in ER3 is exactly the same as described in bug report. After performing  /host=master:reload with the corrupted host.xml server process will quit with ConfigurationPersistenceException.  

Changing status back to assigned

Comment 9 Brian Stansberry 2013-04-01 15:07:36 UTC
See my comment on Dec 10 on AS7-5976:

"The critical task here is to validate the behavior is correct if the reload operation includes the restart-servers=false parameter. In that case the user's intent is the servers remain unaffected by the reload, so having the PC terminate due to failure is incorrect.

Having the PC terminate due to failure to reload the HC is ok if no servers are running, as a PC with no servers and an unusable HC is of little value."


So, use /host=master:reload(restart-servers=false) to check the behavior if you don't want servers affected.

Comment 10 Petr Kremensky 2013-04-04 10:23:03 UTC
Verified on EAP 6.1.0 ER4. 

Corrupting the host.xml and calling /host=master:reload(restart-servers=false) won't affect running servers. PC without running servers is terminated.


Note You need to log in before you can comment on or make changes to this bug.