Fisheye upgrade guide
Fisheye releases
On this page
In this section
Related content
- No related content found
This page describes how to upgrade to a new version of Fisheye. We strongly recommend that you update Fisheye by performing the steps below.
Note that:
- This update process does not perform an in-place upgrade, but installs the new version of Fisheye into a fresh installation directory. The new Fisheye uses your existing Fisheye install directory and external database.
- For production environments we recommend that you test the Fisheye update on a QA server before deploying to production.
- You can update from any previous version to the latest version of Fisheye.
If you want to migrate Fisheye, as well as upgrade, see Migrating and Upgrading Fisheye/Crucible.
1. Review the upgrade notes
There are specific upgrade notes further down this page for each version of Fisheye.
You should read the relevant sections for each version between your current version of Fisheye and the version you are upgrading to.
2. Backup your Fisheye data
Back up your entire Fisheye instance (see Backing up and restoring Fisheye data):
- If you are backing up your Fisheye instance using the Admin interface, tick all of the 'Include' checkboxes (e.g. repository and application caches, plugins and their configuration data, SQL database, etc).
- If you are backing up your Fisheye instance using the command-line interface, do not use any exclusion options.
3. Stop Fisheye
In a terminal, change directory to the <FishEye install directory>
and run this:
bin/stop.sh
4. Install Fisheye
This update process does not perform an in-place upgrade, but installs the new version of Fisheye into a fresh install directory. The new Fisheye uses your existing Fisheye data (FISHEYE_INST
) directory and external database.
Download the Fisheye zip file. On Windows, download the installer. See Installing Fisheye on Linux and Mac or Installing Fisheye on Windows for detailed installation instructions.
FISHEYE_INST
directory (i.e. "Fisheye instance" directory).- The
FISHEYE_INST
directory is the Fisheye data directory (not the installation directory) and has a location defined by theFISHEYE_INST
environment variable. It is used to keep the Fisheye data completely separate from the Fisheye/Crucible application files. We recommend that you configure Fisheye/Crucible to use aFISHEYE_INST
directory for production instances. Read more aboutFISHEYE_INST
in Installing Fisheye on Windows or Installing Fisheye on Linux and Mac. - The
<FishEye install directory>
Method 1: Using a FISHEYE_INST directory
If you have Fisheye/Crucible configured to use a FISHEYE_INST
directory, then follow the instructions below. This is the recommended scenario for production installations.
- Shut down your existing Fisheye/Crucible server, using
bin\stop.bat
orbin\stop.sh
from the<FishEye install directory>
. - Make a backup of your
FISHEYE_INST
directory. - Download Fisheye or Crucible.
- Extract the new Fisheye/Crucible version to a new directory.
- Leave your
FISHEYE_INST
environment variable set to its existing location. Both Fisheye and Crucible use this variable.- Please be aware that jar files in the
FISHEYE_INST/lib
directory may conflict with those required for Fisheye's normal operation. Jar files in this directory should be limited to those which provide functionality not provided by Fisheye (e.g. database drivers).
- Please be aware that jar files in the
- Start Fisheye/Crucible from the new installation directory by running
bin/run.sh
. (Userun.bat
on Windows.) - Follow any version-specific instructions found in the Fisheye upgrade guide or Crucible upgrade guide.
Method 2: Without a FISHEYE_INST directory
If you do not have Fisheye/Crucible configured to use a FISHEYE_INST
directory and do not want to set one up, then follow the instructions below. The <FishEye install directory>
is the location of the existing Fisheye/Crucible installation. Note that this is the typical scenario for evaluation installations, and is not recommended for production installations.
You will need to copy some files from your old Fisheye/Crucible installation to your new one.
- Download Fisheye or Crucible.
- Extract the new Fisheye/Crucible archive into a directory such as
<New FishEye install directory>
. - Shut down the old Fisheye/Crucible server, using
bin\stop.bat
orbin\stop.sh
from the<FishEye install directory>
. - Copy
<FishEye install directory>
<New FishEye install directory>
- Delete the following directories from the
<New FishEye install directory>
<New FishEye install directory>/var/cache
<New FishEye install directory>/var/data
<New FishEye install directory>/var/log
<New FishEye install directory>/data
(If it exists)
- Copy (or move) the following directories from
<FishEye install directory>
<New FishEye install directory>
<FishEye install directory>/var/cache
<FishEye install directory>/var/data
<FishEye install directory>/var/log
<New FishEye install directory>/data
(If it exists)
<FishEye install directory>/var/osgi-cache
<FishEye install directory>/var/plugins
<FishEye install directory>/var/tmp
- Delete the
<New FishEye install directory>
.
- Copy (or move) the
<FishEye install directory>
<New FishEye install directory>
- Start Fisheye/Crucible from the new installation by running
<New FishEye install directory>
/bin/run.shrun.bat
on Windows.) - Follow any version-specific instructions found in the Fisheye upgrade guide or Crucible upgrade guide .
Method 3: Without a FISHEYE_INST directory, but would like to set one up
If you do not have Fisheye/Crucible configured to use a FISHEYE_INST
directory but would like to set one up, then follow the instructions below. You may wish to do this when reconfiguring an existing installation for a production environment.
The FISHEYE_INST
directory is the Fisheye data directory, which has a location defined by the
FISHEYE_INST
environment variable, and which should be completely separate from the <FishEye install directory>
. The <FishEye install directory>
is the location of the existing Fisheye/Crucible installation.
- Download Fisheye or Crucible.
- Shut down the existing Fisheye/Crucible server, using
bin\stop.bat
orbin\stop.sh
from the<FishEye install directory>
. - Set up the
FISHEYE_INST
environment variable, then create theFISHEYE_INST
directory on your file system. - Copy
<FishEye install directory>
/config.xmlFISHEYE_INST
directory. - Copy the
<FishEye install directory>
/varFISHEYE_INST
directory. - Copy the
<FishEye install directory>
/cacheFISHEYE_INST
directory. - If it exists, copy the
<FishEye install directory>
/dataFISHEYE_INST
directory. - Extract the new Fisheye/Crucible archive into a directory such as
<New FishEye install directory>
- Start Fisheye/Crucible from the new installation by running
<New FishEye install directory>
run.bat
on Windows.)
- If your configuration is not automatically picked up and you cannot see your existing repositories, check your Administration > Sys-Info page, where you will see information about the
<FishEye install directory>
andFISHEYE_INST
. Check that yourFISHEYE_INST
is pointing to the right directory.
- If your configuration is not automatically picked up and you cannot see your existing repositories, check your Administration > Sys-Info page, where you will see information about the
- Follow any version-specific instructions found in the Fisheye upgrade guide or Crucible upgrade guide .
Method 4: Using the Fisheye Installer for Windows
If you have Fisheye/Crucible configured to use a FISHEYE_INST
directory, then follow the instructions below. This is the recommended scenario for production installations. This also applies when previously running Fisheye as a Windows Service and the FISHEYE_INST location is defined within the wrapper.
- Shut down your existing Fisheye/Crucible server, using bin\stop.bat or bin\stop.sh from the
<FishEye install directory>
.
If you currently run Fisheye or Crucible as a Windows service, see Upgrading Fisheye on Windows for instructions on uninstalling the service. - Make a backup of your FISHEYE_INST directory.
- Download Fisheye or Crucible.
- Run the installer.
- Select the option to use the Custom Install. Proceed with the install.
- Select the folder where Fisheye will be installed. In thisstepyou can leave the default value selected. Proceed with the install.
- Select the folder where your data will be stored, which in this case is the folder pointed by your FISHEYE_INST. Ensure to select the correct folder in order to have all your data being read by Fisheye. For example, if your FISHEYE_INST is currently set to
C:\Atlassian\fish_inst
, this is the folder you have to select. Proceed with the install until the end.- Please be aware that jar files in the
FISHEYE_INST/lib
directory may conflict with those required for Fisheye's normal operation. Jar files in this directory should be limited to those which provide functionality not provided by Fisheye (e.g. database drivers).
- Please be aware that jar files in the
- Fisheye will be installed and running as a service by the end of the install process.
- Follow any version-specific instructions found in the Fisheye upgrade guide or Crucible upgrade guide.
If you do not have Fisheye/Crucible configured to use a FISHEYE_INST
directory, the installer will create a directory for you and will create a FISHEYE_INST environment variable pointing to it. In the following instructions, <FishEye install directory>
refers to the location of the previous Fisheye/Crucible installation. This also applies when previously running Fisheye as a Windows Service and there is no FISHEYE_INST defined.
- Shut down your existing Fisheye/Crucible server, using bin\stop.bat or bin\stop.sh from the <Fisheye install directory>.
If you currently run Fisheye or Crucible as a Windows service, see Upgrading Fisheye on Windows for instructions on uninstalling the service. - Download Fisheye or Crucible.
- Run the installer.
- Select the option to use the Default Install. You may also use Custom Install if you want to select the directories where Fisheye will be installed and where your data will be stored. Proceed with the install.
- Fisheye will be installed and running as a service by the end of the install process.
- Stop Fisheye service (names as Atlassian Fisheye).
- Copy
<FishEye install directory>
<FISHEYE_INST>
- Delete the following directories from the
<
>FISHEYE_INST
<
>/var/cacheFISHEYE_INST
<
>/var/dataFISHEYE_INST
<
>/var/logFISHEYE_INST
<FISHEYE_INST>/data
(If it exists)
- Copy (or move) the following directories from
<FishEye install directory>
:<
>FISHEYE_INST
<FishEye install directory>/var/cache
<FishEye install directory>/var/data
<FishEye install directory>/var/log
<FishEye install directory>/data
(If it exists)
DO NOT include the following directories when you do that:<FishEye install directory>/var/osgi-cache
<FishEye install directory>/var/plugins
<FishEye install directory>/var/tmp
- Delete the
<
>FISHEYE_INST
.
- Copy (or move) the
<FishEye install directory>
<
>FISHEYE_INST
- Start the Atlassian Fisheye service.
- Follow any version-specific instructions found in the Fisheye upgrade guide or Crucible upgrade guide.
5. Update any custom configurations
Once the new vesion of Fisheye is installed, remember to update any custom configurations in the new version of Fisheye, for example your SQL driver and your wrapper.config file.
If you are using MySQL, read about the JDBC driver.
If you currently run Fisheye as a Windows service and are installing the new version of Fisheye in a new location, you need to uninstall and then reinstall Fisheye as a Windows service. Please see Upgrading Fisheye on Windows for instructions.
For Fisheye 3.4 and later, on Windows, you can edit the Java VM properties using the tool included in the download. See JVM system properties.
6. Start Fisheye
In a terminal, change directory to the <FishEye install directory>
and run this:
bin/start.sh
After a few moments, in a web browser on the same machine, go to http://localhost:8060/
(or, from another machine, type
http://hostname:8060/
, where hostname
is the name of the machine where you installed Fisheye).
Note that the first time you run a new version of Fisheye, it will automatically upgrade its data.
If you are interested in suggestions for improving Fisheye performance see Best Practices for Fisheye configuration and Tuning Fisheye performance.
Version-specific update notes
This section provides specific update notes for each version of Fisheye. These notes supplement the primary update guide above.
You should read the relevant sections for each version between your current version of Fisheye and the version you are upgrading to.
New JVM properties
Two new properties to control in-memory blame cache have been added:
- fisheye.blame.scm.fallback.cache.timeout.in.seconds=N (seconds since adding to cache)
- fisheye.blame.scm.fallback.cache.size=N (number of files)
REST API changes
All review-related endpoints (such as /rest-service/reviews-v1, /rest-service/search-v1 or /rest-service-fecru/recently-visited-v1) can handle multiple issue keys now. The <linkedIssues> is new property to link issues. The <jiraIssueKey> is deprecated, on GET it returns a first issue (alphabetically) from the <linkedIssues> list, on POST it links an issue with a review.
Java API changes
ReviewData(String, String, String, UserData, UserData, UserData, String, String, ReviewData$State, Boolean, PermId, Date, Date, Date, int, String) has been removed
ReviewData(String, String, String, UserData, UserData, UserData, String, String, ReviewData$State, ReviewData$ReviewType, Boolean, PermId, Date, Date, Date, int, String) has been removed
com.atlassian.fisheye.plugin.web.helpers.Helper#getLinkedIssue method has been added
getWebPanelsForIssueInReview(HttpServletRequest, HttpServletResponse, String, String, String) has been added for handling issue transitions in a close review dialog - in case your plugin uses 'crucible.review.complete.dialog.jira.action' or 'crucible.review.close.dialog.jira.action' webPanelKey then it requires an update
Mercurial 4.5 and Git 2.16-2.18 are supported
Crucible 4.6 supports Mercurial up to version 4.5 and Git up to version 2.18.
Git 2.16.x is incompatible with OpenSSH older than 6.8, more details in FE-7016.
Crucible 4.4 upgrade notes
Mercurial 4.1 and Git 2.12 are supported
Crucible 4.4 supports Mercurial 4.1 and Git 2.12 clients
Mercurial 1.5 - 1.8.4 is no longer supported
As of Crucible 4.4, the oldest Mercurial version supported is 1.9.3 (released in 2011). Before you upgrade to Crucible 4.4, upgrade Mercurial client binaries.
Crucible 4.2 upgrade notes
Support for repository renaming
For Crucible 4.2, and later versions, each repository is now identified by both a (display) name and a key. The name can be changed, even for existing repositories, while the key can never be changed. Previously, repositories were identified only by an immutable 'name' attribute (equivalent to the 'key' attribute in Crucible 4.2). See Renaming a repository for more details.
When upgrading to Crucible 4.2, each repository's 'name' will be used to populate both its (display) name and key.
Crucible 4.1 upgrade notes
Subversion 1.9 is supported
Crucible 4.1 supports Subversion 1.9 and the new FSFS format 7 introduced by this release. You may add new Subversion 1.9 repositories to Crucible as well as upgrade (with 'svnadmin upgrade' command) repositories that you've already added. Subversion 1.9 support covers both the bundled SVN client (SVNKit) as well as the Subversion 1.9 native library setup.
Please note however that during internal performance tests we've observed some indexing slowdowns when SVN 1.9 (FSFS format 7) repositories were accessed with the file:// protocol using the bundled SVN client (comparing to SVN 1.8 ones). Therefore, if indexing time is a priority for you, we recommend that you continue to use SVN 1.5 - 1.8 repositories, if possible.
Subversion 1.1 - 1.4 is no longer supported
As of Crucible 4.1, the oldest Subversion version supported is 1.5 (released in 2008). This applies also to the repository format: the lowest supported is format 5 and FSFS format 3. Therefore, before you upgrade to Crucible 4.1, please upgrade both the Subversion client and server binaries as well as your repositories (using the 'svn upgrade' command).
Subversion merges are supported
Crucible 4.1 supports merge operation in Subversion. You may merge branches in Subversion and see information about this in commit graph and on changeset page. More information is available on official documentation page.
Please note that merge information is processed only for new changesets. The repository history does not rescanned in order to expose the merge information for old changesets, however, if you need such historical information, you can perform a full repository reindex.
Crucible 4.0 upgrade notes
User directories migration
Crucible 4.0 includes new user directories administration. Please refer to Fisheye 4.0 user directories migration for details of how users-related settings are migrated.
API changes for plugin developers
Please note that if you're not a plugin developer, changes described in this section will not affect you.
PLUGINS SYSTEM 4.0 UPGRADE
Crucible 4.0 ships with an upgraded Plugins System. Please refer to the Atlassian Plugins 4.0 Upgrade Guide for a detailed list of changes introduced by the upgrade.
CRUCIBLE API CHANGES
Crucible API changes in Crucible 4.0 are mainly related to the introduction of user directories. Host-based authentication has been removed, and LDAP-based authentication is now handled by Crowd rather than directly by Crucible, resulting in the removal of the HostAuthSettings and LdapAuthSettings classes. You could implement a custom authenticator for Crucible or a custom directory connector for Crowd, if absolutely necessary, but we strongly recommend that you rely on existing authentication methods.
The UserService# getActiveUserCount method is deprecated - use the getLicensedUserCount instead.
You can no longer call UserService# setCrucibleEnabledForUsers to enable Crucible access for certain users, as this method has been removed. Instead, you should assign users to a proper group and set group permissions using the new GlobalPermissionService# setPermissionsForGroup method.
The REST API has minor changes related to user and group management. See the documentation for more details.
Dropping Host-based Authentication
As already announced, host-based authentication is no longer supported by Crucible 4.0 and later versions.
Crucible 3.10 upgrade notes
Secure connections using self-signed certificates may fail
Crucible 3.10 uses an updated version of commons-httpclient
that provides SNI (Server Name Indication) support. However, this version of commons-httpclient
uses stricter domain name verification, so webhooks and application links that use a secure connection may stop working. This is only likely to happen when Crucible accesses an application using a secure connection verified by a self-signed certificate, where the application domain name (for example 'jira.company.com') does not match the certificate common name (such as 'company.com').
You may need to update the SSL certificates you use with secure connections between Crucible 3.10 and other applications – self-signed certificates may no longer work as before.
Uppercase project keys upgrade task
The project key upgrade task runs on startup after upgrading to Crucible 3.10; it should only take a few seconds to run.
This is required because Crucible 3.10 now only allows uppercase project keys. We've made this change so that the DB project key column can be properly indexed, allowing much improved performance for a number of different method calls.
- All existing project keys will be converted to be all uppercase and all unique. For Crucible, this also includes the project key part of review keys.
- If a project key conflict occurs, the project key will be renamed by adding an
UPGRADE[Number]
suffix. You can change renamed project keys manually after the upgrade if necessary. - The upgrade task produces logs for project key changes. Look for logs starting with
[projectKey.uppercase]
- All operations using the project key as an argument are case sensitive, with the exception that view operations in the browser are case insensitive and will upper case the project key automatically.
- Fisheye and Crucible track recently visited projects, reviews and snippets for every user. Any projects, reviews and snippets with renamed project keys however will not appear in the recently visited cache.
If there are entity links between Jira Software and Crucible projects, the mapping will be automatically updated. You can check this by visiting
Administration > Add-ons > Fisheye configuration > Fisheye/Crucible entity mappings in Jira. If Crucible projects are not all uppercased, click Refresh cache to update the mapping.
Note that Crucible will refuse to start if the DB is not case-sensitive with UTF-8 default encoding, to avoid potential data corruption during the upgrade task. In this case you'll see the following log message:
FeCru connecting to DB not using case-sensitive UTF8 encoding
LDAP synchronization
Crucible 3.10 now supports paging (with a default page size of 1000) when requesting data from the LDAP server, and works seamlessly when the number of user accounts exceeds 1000 in Active Directory. It reverts to the previous behavior with servers that don't support LDAPv3.
The paging size can be controlled with the fisheye.ldap.sync.page.size
system property. Setting it to 0 disables paging.
Crucible 3.9 upgrade notes
Lucene index upgrade task
A reindex of the Lucene index runs on startup after upgrading to Crucible 3.9. This is required because new fields have been added to the Lucene index to enable sorting of reviews by the Review column ( CRUC-6560 - Review dashboard can't be sorted by Review key CLOSED ).
Reindexing takes about 10 minutes for 15000 reviews. During reindex, Crucible is usable but some reviews may not be visible.
Smart Commits now use the Jira REST API
As of version 3.9.0, Crucible uses the Jira REST API instead of the Jira Fisheye Plugin for Jira Smart Commits. This change ensures that you will be able to use Smart Commits with future versions of Jira. Note that using Smart Commits with previous versions of Crucible (earlier than 3.9.0) and Jira 7 will cause an error notification via email.
Git clone changes
As of version 3.9.0, Crucible turns Git garbage collection off when cloning a repository (by adding the gc.pruneExpire=never
option) to prevent unreferenced objects being removed from local clones. Also, when cloning a repository, git config is run on each repository during instance startup. This change bumps the Git cache version ( CACHE_VERSION
) to 22.
Supported platform upgrades
- Oracle 12c is now supported.
- Git 2.4.6 and Mercurial 3.4.2 are now supported.
- Support for Java 7 has been removed from Crucible 3.9, as previously announced.
- Support for Internet Explorer 9 has been removed from Crucible 3.9, as previously announced.
Crucible 3.8 upgrade notes
New database index to improve Review Dashboard and Review Search pages load time
A database upgrade task is run on startup that adds a database index. This task should take no more than a few minutes, and should complete within seconds on most instances.
Improved Git indexing time for newly created branches
In order to allow faster Git indexing, a new field was added to the internal repository caches. The cache for each Git repository is automatically upgraded when the repository is started for the first time after upgrading. It is expected to take under a minute per repository, but may take slightly longer if repositories have thousands of branches.
Supported platform upgrades
- Java 7 is deprecated, as previously announced.
Crucible 3.7 upgrade notes
Supported platform upgrades
- Java 7 support is deprecated – please see the End of Support Announcements for Crucible.
Crucible 3.6 upgrade notes
Supported platform upgrades
- Java 6 is not supported by Crucible 3.6, as previously announced.
- Java 7 support is deprecated – please see the End of Support Announcements for Crucible .
SSLv3 support disabled by default
As of Crucible 3.6, SSLv3 support is disabled by default. This shouldn't affect normal operations in supported browsers. If you need to re-enable SSLv3 support, please consult Configuring SSL cipher suites for Jetty.
Crucible 3.5 upgrade notes
An upgrade from an earlier version of Crucible to 3.5.0 may cause problems if you have upgraded the Universal Plugin Manager Plugin to a newer version than is shipped with Crucible 3.5.0.
The workaround for this is to remove the custom installed version of the Universal Plugin Manager Plugin.
After upgrading from 3.4.5 to 3.5.0, this error is printed in the web browser when you try to access some pages:
javax.servlet.jsp.JspException: javax.el.ELException: java.lang.NullPointerException: couldn't locate WebResourceIntegration service
Workaround:
- Stop the new Crucible instance;
- Remove your newer version of the Universal Plugin Manager Plugin at $FISHEYE_INST/var/plugins/user/plugin.xxxxxx.atlassian-universal-plugin-manager-plugin*.jar;
- Start the new Crucible instance again.
Crucible 3.4 upgrade notes
Windows installer
We've produced 32-bit and 64-bit installers for Crucible on Windows. Each installer adds Crucible as a Windows service, and starts the service, automatically. The express install creates, by default, a Data
directory and a separate install directory in C:\Atlassian
. The custom install mode allows you to choose different locations for the install and Data
directories, with the restriction that the Data
directory must not be contained in the install directory. The installer creates the FISHEYE_INST
system environment variable.
See Installing Crucible on Windows for detailed installation instructions.
Download the Crucible installer here.
Crucible may now bind to a different IP address on Windows
Prior to Crucible 3.4, a bug in Crucible ( FE-4909 CLOSED ) meant that Crucible may not have correctly bound to the IP address you configured. This may have happened if you configured Crucible to bind to a single IP address on a network interface that has several IP addresses; Crucible may in fact have bound to a different IP address. For example, if you have an interface with the IP addresses 1.2.3.4 and 1.2.3.5, and you configured Crucible to use 1.2.3.5, it may have incorrectly bound to 1.2.3.4.
Now that the bug is fixed, Crucible 3.4, and later versions, will now correctly bind to the configured IP address, although this may now be different from the previously bound address.
v1 REST API resources deprecated
Note that the 'v1' REST API resources are deprecated and will be removed in a future release. See the Fisheye Crucible REST API.
Crucible 3.3 upgrade notes
The following platforms are no longer supported by Crucible 3.3:
- Internet Explorer 8
- MySQL 5.0
- PostgreSQL 8.2
- SQL Server 2005
Please read the End of Support Announcements for Crucible.
Supported platform upgrades
- SVN 1.8 is supported by Crucible 3.3.
- Microsoft Internet Explorer 11 is supported by Crucible 3.3.
See the Crucible Supported platforms.
Crucible 3.2 upgrade notes
REST endpoint change
For Crucible 3.2, the RestReviewService.remindIncompleteReviewers() ('/rest-service/reviews-v1/{reviewId}/remind') end point was changed from accepting 'application/x-www-form-urlencoded' content type with 'message' and 'recipient' params to 'application/json' content type with 'message' and 'recipients' JSON fields. See https://docs.atlassian.com/fisheye-crucible/latest/wadl/crucible.html#d2e1881.
User data is moved from data0.bin to the SQL database
An upgrade task is run on startup that moves user data to the SQL database. We are doing this to mitigate the risk of data corruption or loss.
XSRF protection
Crucible 3.2.0 includes protection against XSRF attacks.
If you have implemented a LightSCM plugin, and have used the SimpleConfigurationServlet
base class provided in the scmutils
library, you will need to modify your administration page so that it performs 'delete' operations using an HTTP POST, not a GET.
You can have Crucible convert your anchor tags to form POSTs at runtime by giving them the class "anchor-post". For example, the anchor:
<a class="anchor-post" href="#" data-href="./fsscm?name=TEST&delete=true">Delete</a>
will be converted into a form which POSTs to ./fsscm with form parameters name=TEST and delete=true.
Internally managed Git repositories no longer supported
As previously announced, internally managed Git repositories are no longer supported by Crucible 3.2.
Please read the migration guide for information about options and procedures for migrating your internally managed Git repositories – note that we recommend that you upgrade to Crucible 3.2 before migrating any internally managed repositories.
Supported platform upgrades
- Communication with Jira versions older than 5.0 is no longer supported.
- Microsoft SQL Server 2012 is now supported (support for SQL Server 2005 is deprecated).
- Microsoft Internet Explorer 10 is now supported (support for IE 8 is deprecated).
- MySQL 5.0 is deprecated.
- PostgreSQL 8.2 is deprecated.
- The Atlassian AUI plugin has been upgraded to AUI 5.2.
- jQuery has been upgraded to 1.8.3.
- backbonejs has been upgraded to 1.0.
Crucible 3.1 upgrade notes
Crucible 3.1 Merge some per-repository Lucene indices into a global cross-repository Lucene index
Crucible 3.1 has greatly improved performance and scalability for QuickSearch and QuickNav. To achieve this, the per-repository 'METADATA' Lucene indices will be moved into a single global cross-repository Lucene index. This means Crucible is able to search across more repositories in less time because now only a single search index needs to be queried instead of the previous N. Merging these indices into the single cross-repository index can be refreshed in two ways:
- Recommended: As an upgrade task that is run automatically when Crucible 3.1 is run for the first time.
- As an offline process on a separate staging server.
During the automatic upgrade task, Crucible is fully usable and functional, although search results for files, commits and committers may be incomplete.
In our testing we have found that the automatic upgrade task took 4 hours to complete on a Crucible instance with 144 repositories of different kinds, with 58 GB of data in the FISHEYE_INST folder (excluding logs). We are confident that the automatic upgrade task is suitable for the majority of production Crucible installations. It is worth repeating that the instance was fully functional (reviews, Jira Integration, Activity Streams, Charts etc) apart from Quick Nav and Quick Search during this time.
Nevertheless, where required, we provide instructions for performing the reindex as an offline process on a separate staging server.
Plugin Settings will be moved from the config.xml to the SQL database
As of Crucible 3.1.0, plugin settings which were previously stored in the <properties> element inside config.xml will be stored in the SQL database. This includes settings for any bundled plugins such as ApplicationLinks, the UniversalPluginManager etc, and any 3rd party plugins.
An upgrade task is run on startup which will first insert all the properties found in config.xml into a new table in the SQL database. Once successful, the properties are removed from config.xml.
As part of this change, the RepositoryOptions.setProperties (Map<String, String>properties) and RepositoryOptions.getProperties() methods have been removed from our API. If you are using a plugin which uses either of these methods, you will need to update the plugin to a version which uses the Spring component PluginSettingsFactory. Plugins can use this to access the migrated global and per-repository properties that were previously available via the old RepositoryOptions API.
Crucible 3.0 upgrade notes
Jetty 8
Crucible 3.0 now uses Jetty 8 as its web server and Java servelet container. This change should be completely transparent when you upgrade to Crucible 3.0. However, if you have customised either your jetty-web.xml
file, or the maxFormContentSize
system property, you will need to update those in the new version. See Enabling Access Logging in Fisheye and this Fisheye KB page for more information.
Infinity DB
Crucible 3.0 now uses the InfinityDB 3.0 database internally to provide improved performance for concurrent access to Crucible. This change is transparent to users in all respects.
Pipelined indexing
Crucible 3.0 introduces a new indexing approach that splits the repository indexing process into separate tasks that can be performed in a phased and concurrent way. Users will benefit from the way in which Crucible functionality, such as review creation, now becomes available as indexing progresses. This change is transparent to users in all other respects. See Pipelined indexing.
Improved handling of user preferences with session cookies
Upgrading may result in some users losing their preferences.
SQL Server transaction isolation configuration
We recommend a configuration change for SQL Server to use snapshot mode for the transaction isolation level – see Migrating to SQL Server. This change avoids occasional database deadlocks, and prevents performance warning messages in the logs and admin screens.
Checking for known issues and troubleshooting the Fisheye upgrade
If something is not working correctly after you have completed the steps above to upgrade your Fisheye installation, please check for known Fisheye issues and try troubleshooting your upgrade as described below:
- Check for known issues. Sometimes we find out about a problem with the latest version of Fisheye after we have released the software. In such cases we publish information about the known issues in the Fisheye Knowledge Base. Please check the Fisheye and Crucible Known Issues in the Fisheye Knowledge Base and follow the instructions to apply any necessary patches if necessary.
- Did you encounter a problem during the Fisheye upgrade? Please refer to the guide to troubleshooting upgrades in the Fisheye Knowledge Base.
- If you encounter a problem during the upgrade and cannot solve it, please create a support ticket and one of our support engineers will help you.
In this section
Related content
- No related content found