4. Fixed Bugs in JBoss Operations Network 3.0.1
These bugs have been fixed in JBoss Operations Network 3.0.1. For a complete list of all issues addressed in JBoss Operations Network 3.0.1, search Bugzilla for RHQ Project bugs with a target of JON 3.0.1.
| Bug Number | Component | Description |
|---|---|---|
| Bugzilla 785985 | Agent | The agent was not properly updating its configuration if the configuration changed in the server while the agent was offline. |
| Bugzilla 772514 | CLI | Extracting the JBoss ON CLI archive resulted in a world-writable installation directory for the CLI. |
| Bugzilla 785773 | CLI | In 3.0, the CLI improperly reset the exporter variable after each call. The exporter variable should have persisted. |
| Bugzilla 790074 | CLI | If an argument passed to the CLI had a space in it, each space-separated word was treated as a separate argument. The CLI now properly interprets arguments with spaces. |
| Bugzilla 784361 | CLI | A change to the DiscoveryBossBean class broke discovery scans that were initiated through CLI scripts. |
| Bugzilla 789014 | CLI | The updateBackingContent method did not allow a version number to be assigned to a package; it automatically incremented whatever the previous installed version was. This was sub-optimal in the assigned version number and it made it difficult to have a consistent versioning scheme in both the UI and the CLI. |
| Bugzilla 788629 | CLI | The updateBackingContent method did not properly persist uploaded files. The version for the backed content was not properly maintained, so after an initial upload and then an update, the package was not updated. |
| Bugzilla 765795 | CLI | The createPackageBackedResource method threw a stack overflow exception. |
| Bugzilla 789018 | CLI | The retrieveBackingContent and updateBackingContent methods returned a null pointer exception if multiple installed packages were attached to a web app. A web app is only allowed to have a single installed package, but a database error could result in multiple packages being attached to the web app simultaneously. The agent now accepts the content update operation and corrects any problems during discovery. |
| Bugzilla 789027 | Content | The disk source content provider was originally designed only to import individual files. If a parent folder contained a subfolder, the disk source sync operation failed with a FileNotFoundException error. |
| Bugzilla 785022 | Database (Upgrade) | The database being upgraded already had an index file, then the db-upgrade.xml task failed to execute and upgrade failed. |
| Bugzilla 785968 | Drift | A drift template could not be edited if more than one drift definition was attached to it. |
| Bugzilla 761320 | Drift | The drift definition schedule could get pushed out, so that the detection scan never runs for that definition. Every time an agent scan ran, it updated both the agent scan interval and the drift run interval; since the agent scan runs more frequently, it was pushing the drift run schedule out so that drift never ran. |
| Bugzilla 789454 | Drift | If a drift directory had a Unix special file in it which drift could not process, such as a named pipe, then the drift detection failed for the entire directory. |
| Bugzilla 789492 | Drift | The backslash in Windows directory path names was misinterpreted as escaping the character in the path name. This resulted in incorrect or non-existent paths for drift directories on Windows. |
| Bugzilla 786613 | Drift | The drift detection process fetched all files, recursively, under the base directory to check for files that match the filters. (The drift process did not digest all the files, but it did check them.) If the base directory were set to a root or a broad directory directory, then effectively the agent scanned the entire filesystem to check for matching files. The agent could hang or have very poor performance. |
| Bugzilla 785847 | Drift | When the server requested content for files which had previously been detected, it also sent the agent a list of SHAs to identify the content. The agent verifies the content by reading the snapshot file. The agent would read, close, and re-open the snapshot file for every SHA on the list, creating excessive I/O. |
| Bugzilla 786611 | Drift | When diffing files between snapshots, the file selected first was treated as the older file and the file selected second was treated as the newer version, regardless of the actual timestamp of the files in the snapshot. Now, the newer file is always treated as the newer version in the diff, regardless of what order the files are selected when running the diff. |
| Bugzilla 785857 | Drift | If a template was deleted, any resource with detached definitions based on that template would throw an exception and none of the definitions for that resource could be listed. |
| Bugzilla 784730 | Events | When the agent was initially started, the logfile event pollers were blocked because of a write lock on the managers during the plug-in container initialization. The lock lasted longer than the poller creation timeout period, so creation would silently fail. This meant that event logs were not collected because the pollers weren't running. (The pollers would be created some time after agent startup when the plug-in container attempted to connects a component.) |
| Bugzilla 785026 | Events | If event logging were enabled but native support disabled, events were not processed. |
| Bugzilla 781621 | LDAP Authentication | Configuring LDAP authentication over an SSL connection failed with a ServiceUnavailableException error. |
| Bugzilla 790079 | Provisioning | Redeploying a bundle with the clean option would delete any files which had been edited instead of replacing them with the version in the bundle. |
| Bugzilla 786603 | Provisioning | Deploying a bundle with the manageRootDir parameter missing or set to true in the recipe file (the default) would not delete the files in the destination directory, so the destination directory would not match the bundle structure. |
| Bugzilla 789027 | Plug-ins (JBoss AS) | The discovery process did not use the home directory of a JBoss instance as a relative path for any children scripts it discovered; rather, it used the home directory of the agent's installation directory. When the agent attempted to compute a SHA-256 hash for the file after it was discovered, it could not find the file in the (wrong) location and returned a FileNotFoundException. |
| Bugzilla 790908 | Plug-ins (JBoss AS) | Some of the properties and metadata stored for JBoss child resources could be very large. The JBoss AS 5 resource could end up consuming so much memory that it caused out of memory errors on the agent. Unused properties for JBoss AS 5 child resources have been filtered to reduce the memory use. |
| Bugzilla 773661 | Plug-ins (JBoss AS) | By default, the JMX console in EAP is secured, so the agent could not connect to the JMX server to retrieve required information. However, the connection attempt failed silently. Instead of reporting that the connection attempt failed, the agent reported that the JMX server was unavailable. |
| Bugzilla 790075 | Plug-ins (JBoss AS) | EAR and WAR child resources had some plug-in configuration properties that were unset by default yet were required. This mean that opening the plug-in configuration page for the resource would immediately cause an error. |
| Bugzilla 771201 | Plug-ins (JBoss EPP) | The detection run for EPP 5.2 resources would fail with a StringIndexOutOfBoundsException. This prevented an EPP instance from being added to the inventory. |
| Bugzilla 783879 | Plug-ins (Platform) | The FileSystemComponent would return an invalid diskQueue metric the first time, which would then launch a refresh operation which got the correct metric.
Because the metric was being collected twice, this could cause a performance problem on a system where a lot of metrics were collected or multiple filesystems were being monitored.
|
| Bugzilla 736737 | Plug-ins (Tomcat) | If the connection to the Tomcat server fails for any reason, the agent caught the entire exception and simply marked the Tomcat resource as down. Essentially availability checks for Tomcat could fail silently. |
| Bugzilla 771495 | Plug-ins (Tomcat) | The regular expression used to discover Tomcat WAR files could exclude some valid WAR deployments because the way it interpreted the WAR name did not always match the value in the Tomcat virtual host. The regular expression has been fixed. |
| Bugzilla 772761 | Plug-ins (Tomcat) | The Tomcat plug-in was hard-coded to use a one-second timeout value for operations. This caused start, stop, and restart operations to fail routinely. A new operation parameter has been added to allow administrators to set a different timeout value. |
| Bugzilla 788733 | Plug-ins (JBoss AS and Tomcat) | Deleting a content-backed resource (WAR or EAR) and then recreating a new WAR/EAR resource with the same name preserved all of the package history and version information for the deleted web app. |
| Bugzilla 773737 | Server | The version of JBossWeb used by the JBoss ON server has been upgraded to include fixes for CVE-2011-4858, to prevent possible denial-of-service attacks through the web UI. |
| Bugzilla 789419 | UI | Intermittent problems in the UI could interfere with right-click menus. Specifically, when creating a new child resource, not all possible child types may be listed in the right-click menu. |
| Bugzilla 785019 | UI | When LDAP authentication was enabled, if an LDAP user attempting to log into the UI first, they would get only a blank screen and could not log in. If a JBoss ON user logged into the system first, so that the login screen was in the browser cache, then login was successful. |
