Appendix A. The Bugzilla FAQ

If you want to take a test ride, there are test installations at http://landfill.bugzilla.org/, ready to play with directly from your browser.

Bugzilla is covered by the Mozilla Public License. See details at http://www.mozilla.org/MPL/.

http://www.bugzilla.org/support/consulting.html is a list of companies and individuals who have asked us to list them as consultants for Bugzilla.

There are several experienced Bugzilla hackers on the mailing list/newsgroup who are willing to make themselves available for generous compensation. Try sending a message to the mailing list asking for a volunteer.

There are dozens of major companies with public Bugzilla sites to track bugs in their products. We have a fairly complete list available on our website at http://bugzilla.org/installation-list/. If you have an installation of Bugzilla and would like to be added to the list, whether it's a public install or not, simply e-mail Gerv <gerv@mozilla.org>.

A core team, led by Dave Miller (justdave@bugzilla.org).

We can't find any head-to-head comparisons of Bugzilla against other defect-tracking software. If you know of one, please get in touch. In the experience of Matthew Barnson (the original author of this FAQ), though, Bugzilla offers superior performance on commodity hardware, better price (free!), more developer-friendly features (such as stored queries, email integration, and platform independence), improved scalability, greater flexibility, and superior ease-of-use when compared to commercial bug-tracking software.

If you happen to be a vendor for commercial bug-tracking software, and would like to submit a list of advantages your product has over Bugzilla, simply send it to <documentation@bugzilla.org> and we'd be happy to include the comparison in our documentation.

It may be that the support has not been built yet, or that you have not yet found it. While Bugzilla makes strides in usability, customizability, scalability, and user interface with each release, that doesn't mean it can't still use improvement!

The best way to make an enhancement request is to file a bug at bugzilla.mozilla.org and set the Severity to 'enhancement'. Your 'request for enhancement' (RFE) will start out in the UNCONFIRMED state, and will stay there until someone with the ability to CONFIRM the bug reviews it. If that person feels it to be a good request that fits in with Bugzilla's overall direction, the status will be changed to NEW; if not, they will probably explain why and set the bug to RESOLVED/WONTFIX. If someone else has made the same (or almost the same) request before, your request will be marked RESOLVED/DUPLICATE, and a pointer to the previous RFE will be added.

Even if your RFE gets approved, that doesn't mean it's going to make it right into the next release; there are a limited number of developers, and a whole lot of RFEs... some of which are quite complex. If you're a code-hacking sort of person, you can help the project along by making a patch yourself that supports the functionality you require. If you have never contributed anything to Bugzilla before, please be sure to read the Developers' Guide and Contributors' Guide before going ahead.

MySQL was originally chosen because it is free, easy to install, and was available for the hardware Netscape intended to run it on.

Bugzilla 2.20 contains experimental support for PostgreSQL. Bugzilla 2.22 contains complete, stable support for PostgreSQL. As of this release, using PostgreSQL with Bugzilla should be as stable as using MySQL. If you experience any problems with PostgreSQL compatibility, they will be taken as seriously as if you were running MySQL.

Red Hat once ran a version of Bugzilla that worked on Oracle, but that was long, long ago; that version (Bugzilla 2.8) is now obsolete, insecure, and totally unsupported.

In August of 2005, Wim Coekaerts (Director of Linux Engineering at Oracle Corporation) wrote to Dave Miller confirming that Oracle intends to implement and support Bugzilla. Since then, no further information has been forthcoming. Track progress at Bug 189947.

Sybase support is no longer being worked on. Even if it eventually happens, it's VERY unlikely to work without the end-user-company having to stick a few developers on making several manual changes. Sybase is just NOT very standards-compliant (despite all the hype), and it turned out that way too much had to be changed to make it work -- like moving half of the application logic into stored procedures to get any kind of decent performance out of it. Bug 173130 is the relevant bug.

Bug 237862 is a good bug to read through if you'd like to see what progress is being made on general database compatibility.

Bugzilla used to have the path to perl on the shebang line set to /usr/bonsaitools/bin/perl because when Terry first started writing the code for mozilla.org he needed a version of Perl and other tools that were completely under his control. This location was abandoned for the 2.18 release in favor of the more sensible /usr/bin/perl. If you installed an older version of Bugzilla and created the symlink we suggested, you can remove it now (provided that you don't have anything else, such as Bonsai, using it and you don't intend to reinstall an older version of Bugzilla).

The easiest way to get around this is to create a link from one to the other: ln -s /usr/local/bin/perl /usr/bin/perl. If that's not an option for you, the following bit of perl magic will change all the shebang lines (that is to say, the line at the top of each file that starts with '#!' and contains the path) to something else:

perl -pi -e 's@#\!/usr/bin/perl@#\!/usr/local/bin/perl@' *cgi *pl
          

Sadly, this command-line won't work on Windows unless you also have Cygwin. However, MySQL comes with a binary called replace which can do the job:

C:\mysql\bin\replace "#!/usr/bin/perl" "#!C:\perl\bin\perl" -- *.cgi *.pl
          

If your perl path is something else again, just follow the above examples and replace /usr/local/bin/perl with your own perl path.

Once you've modified all your files, you'll also need to modify the t/002goodperl.t test, as it tests that all shebang lines are equal to /usr/bin/perl. (For more information on the test suite, please check out the appropriate section in the Developers' Guide.) Having done this, run the test itself:

perl runtests.pl 2 --verbose

to ensure that you've modified all the relevant files.

If using Apache on Windows, you can avoid the whole problem by setting the ScriptInterpreterSource directive to 'Registry'. (If using Apache 2 or higher, set it to 'Registry-Strict'.) ScriptInterperterSource requires a registry entry "HKEY_CLASSES_ROOT\.cgi\Shell\ExecCGI\Command" to associate .cgi files with your perl executable. If one does not already exist, create it with a default value of "<full path to perl> -T", e.g. "C:\Perl\bin\perl.exe -T".

At present, no.

At present, no. Work is slowly taking place to remove global variables, use $cgi, and use DBI. These are all necessary for mod_perl (as well as being good for other reasons). Visit bug 87406 to view the discussion and progress.

Unfortunately there are no step-by-step instructions, but the following URL contains hints on how to do it: http://fedora.redhat.com/docs/selinux-apache-fc3/sn-debugging-and-customizing.html

It is web and e-mail based.

Yes. However, modifying some fields, notably those related to bug progression states, also require adjusting the program logic to compensate for the change.

There is no GUI for adding fields to Bugzilla at this time. You can follow development of this feature in bug 91037

Yes. Look at https://bugzilla.mozilla.org/report.cgi for samples of what Bugzilla can do in reporting and graphing. Fuller documentation is provided in Section 5.11.

If you can not get the reports you want from the included reporting scripts, it is possible to hook up a professional reporting package such as Crystal Reports using ODBC. If you choose to do this, beware that giving direct access to the database does contain some security implications. Even if you give read-only access to the bugs database it will bypass the secure bugs features of Bugzilla.

Email notification is user-configurable. By default, the bug id and summary of the bug report accompany each email notification, along with a list of the changes made.

Bugzilla email is sent in plain text, the most compatible mail format on the planet.

If you decide to use the bugzilla_email integration features to allow Bugzilla to record responses to mail with the associated bug, you may need to caution your users to set their mailer to "respond to messages in the format in which they were sent". For security reasons Bugzilla ignores HTML tags in comments, and if a user sends HTML-based email into Bugzilla the resulting comment looks downright awful.

Bugzilla can output buglists as HTML (the default), CSV or RDF. The link for CSV can be found at the bottom of the buglist in HTML format. This CSV format can easily be imported into MS Excel or other spreadsheet applications.

To use the RDF format of the buglist it is necessary to append a &ctype=rdf to the URL. RDF is meant to be machine readable and thus it is assumed that the URL would be generated programmatically so there is no user visible link to this format.

Currently the only script included with Bugzilla that can import data is importxml.pl which is intended to be used for importing the data generated by the XML ctype of show_bug.cgi in association with bug moving. Any other use is left as an exercise for the user.

There are also scripts included in the contrib/ directory for using e-mail to import information into Bugzilla, but these scripts are not currently supported and included for educational purposes.

Yes. For more information including available translated templates, see http://www.bugzilla.org/download.html#localizations. Some admin interfaces have been templatized (for easy localization) but many of them are still available in English only. Also, there may be issues with the charset not being declared. See bug 126226 for more information.

Yes. No. Yes (using the CSV format).

MySQL, the database back-end for Bugzilla, allows hot-backup of data. You can find strategies for dealing with backup considerations at http://www.mysql.com/doc/B/a/Backup.html.

If Bugzilla is set up correctly from the start, continuing maintenance needs are minimal and can be done easily using the web interface.

Commercial Bug-tracking software typically costs somewhere upwards of $20,000 or more for 5-10 floating licenses. Bugzilla consultation is available from skilled members of the newsgroup. Simple questions are answered there and then.

It all depends on your level of commitment. Someone with much Bugzilla experience can get you up and running in less than a day, and your Bugzilla install can run untended for years. If your Bugzilla strategy is critical to your business workflow, hire somebody to who has reasonable Perl skills, and a familiarity with the operating system on which Bugzilla will be running, and have them handle your process management, bug-tracking maintenance, and local customization.

No. Bugzilla, Perl, the Template Toolkit, and all other support software needed to make Bugzilla work can be downloaded for free. MySQL -- the database used by Bugzilla -- is also open-source, but they ask that if you find their product valuable, you purchase a support contract from them that suits your needs.

Yes! As of Bugzilla 2.18, it is a simple matter to change the word 'bug' into whatever word/phrase is used by your organization. See the documentation on Customization for more details, specifically Section 6.2.5.

Bugzilla does not lock records. It provides mid-air collision detection -- which means that it warns a user when a commit is about to conflict with commits recently made by another user, and offers the second user a choice of options to deal with the conflict.

Yes, but commits to the database must wait until the tables are unlocked. Bugzilla databases are typically very small, and backups routinely take less than a minute. If your database is larger, you may want to look into alternate backup techniques, such as database replication, or backing up from a read-only mirror. (Read up on these in the MySQL docs on the MySQL site.)

1. Make a backup of both your Bugzilla directory and the database. For the Bugzilla directory this is as easy as doing cp -rp bugzilla bugzilla.bak. For the database, there's a number of options - see the MySQL docs and pick the one that fits you best (the easiest is to just make a physical copy of the database on the disk, but you have to have the database server shut down to do that without risking dataloss).

2. Make the Bugzilla directory your current directory.

3. Use cvs -q update -AdP if you want to update to the tip or cvs -q update -dP -rTAGNAME if you want a specific version (in that case you'll have to replace TAGNAME with a CVS tag name such as BUGZILLA-2_16_5).

   If you've made no local changes, this should be very clean. If you have made local changes, then watch the cvs output for C results. If you get any lines that start with a C it means there were conflicts between your local changes and what's in CVS. You'll need to fix those manually before continuing.

4. After resolving any conflicts that the cvs update operation generated, running ./checksetup.pl will take care of updating the database for you as well as any other changes required for the new version to operate.

   Once you run checksetup.pl, the only way to go back is to restore the database backups. You can't "downgrade" the system cleanly under most circumstances.

See also the instructions in Section 3.14.3.1.

To use the UNCONFIRMED status, you must have the 'usevotes' parameter set to "On". You must then visit the editproducts.cgi page and set the " Number of votes a bug in this product needs to automatically get out of the UNCONFIRMED state" to be a non-zero number. (You will have to do this for each product that wants to use the UNCONFIRMED state.) If you do not actually want users to be able to vote for bugs entered against this product, leave the "Maximum votes per person" value at '0'.

There is work being done to decouple the UNCONFIRMED state from the 'usevotes' parameter for future versions of Bugzilla. Follow the discussion and progress at bug 162060.

Use mysqldump to make a backup of the bugs database. For a typical Bugzilla setup, such a command might look like this:

/usr/bin/mysqldump -u(username) -p(password) --database bugs > bugzilla-backup.txt

See the mysqldump documentation for more information on using the tool, including how to restore your copy onto the destination machine.

Depending on the size of your database, and the power of your machine, the mysqldump command could be running long enough that the password would be visible to someone using the ps command. If you are on a multi-user machine, and this is a concern to you, create an entry in the file ~/.my.cnf that looks like this: [mysqldump] user=bugs password=mypassword and then leave the 'user' and 'password' params out of the command line.

On your new machine, follow the instructions found in Chapter 2 as far as setting up the physical environment of the new machine with perl, webserver, modules, etc. Having done that, you can either: copy your entire Bugzilla directory from the old machine to a new one (if you want to keep your existing code and modifications), or download a newer version (if you are planning to upgrade at the same time). Even if you are upgrading to clean code, you will still want to bring over the localconfig file, and the data directory from the old machine, as they contain configuration information that you probably won't want to re-create.

If the location or port number of your SQL server changed as part of the move, you'll need to update the appropriate variables in localconfig before taking the next step.

Once you have your code in place, and your database has been restored from the backup you made in step 1, run checksetup.pl. This will upgrade your database (if necessary), rebuild your templates, etc.

Run MySQL like this: mysqld --skip-grant-tables. Please remember that this makes MySQL as secure as taping a $100 to the floor of a football stadium bathroom for safekeeping.

This can't be stressed enough. Doing this is a bad idea. Please consult Section 4.2 of this guide and the MySQL documentation for better solutions.

The Bugzilla code has undergone a reasonably complete security audit, and user-facing CGIs run under Perl's taint mode. However, it is recommended that you closely examine permissions on your Bugzilla installation, and follow the recommended security guidelines found in The Bugzilla Guide.

The user can stop Bugzilla from sending any mail by unchecking all boxes on the 'Edit prefs' -> 'Email settings' page. (As of 2.18,this is made easier by the addition of a 'Disable All Mail' button.) Alternately, you can add their email address to the data/nomail file (one email address per line). This will override their personal preferences, and they will never be sent mail again.

To disable email, set the mail_delivery_method parameter to none (2.20 and later), or

$enableSendMail

parameter to '0' in either BugMail.pm (2.18 and later) or processmail (up to 2.16.x).

Up to 2.16.x, changing $enableSendMail will only affect bugmail; email related to password changes, email address changes, bug imports, flag changes, etc. will still be sent out. As of the final release of 2.18, however, the above step will disable all mail sent from Bugzilla for any purpose.

To have bugmail (and only bugmail) redirected to you instead of its intended recipients, leave

$enableSendMail

alone; instead, edit the "newchangedmail" parameter as follows:

• Replace "To:" with "X-Real-To:"

• Replace "Cc:" with "X-Real-CC:"

• Add a "To: %lt;your_email_address>"

For older versions of Bugzilla, you may be able to apply Klaas Freitag's patch for "whineatassigned", which can be found in bug 6679. Note that this patch was made in 2000, so it may take some work to apply cleanly to any releases of Bugzilla newer than that, but you can use it as a starting point.

An updated (and much-expanded) version of this functionality is due to be released as part of Bugzilla 2.20; see bug 185090 for the discussion, and for more up-to-date patches if you just can't wait.

You can find an updated README.mailif file in the contrib/ directory of your Bugzilla distribution that walks you through the setup.

If you are using sendmail, try enabling sendmailnow in editparams.cgi. For earlier versions of sendmail, one could achieve significant performance improvement in the UI (at the cost of delaying the sending of mail) by setting this parameter to off. Sites with sendmail version 8.12 (or higher) should leave this on, as they will not see any performance benefit.

If you are using an alternate MTA, make sure the options given in Bugzilla/BugMail.pm and any other place where sendmail is called are correct for your MTA.

Double-check that you have not turned off email in your user preferences. Confirm that Bugzilla is able to send email by visiting the "Log In" link of your Bugzilla installation and clicking the "Submit Request" button after entering your email address.

If you never receive mail from Bugzilla, chances are you do not have sendmail in "/usr/lib/sendmail". Ensure sendmail lives in, or is symlinked to, "/usr/lib/sendmail".

If you are using an MTA other than sendmail the sendmailnow param must be set to on or no mail will be sent.

Run the "sanity check" utility (sanitycheck.cgi) from your web browser to see! If it finishes without errors, you're probably OK. If it doesn't come back OK (i.e. any red letters), there are certain things Bugzilla can recover from and certain things it can't. If it can't auto-recover, I hope you're familiar with mysqladmin commands or have installed another way to manage your database. Sanity Check, although it is a good basic check on your database integrity, by no means is a substitute for competent database administration and avoiding deletion of data. It is not exhaustive, and was created to do a basic check for the most common problems in Bugzilla databases.

There is no facility in Bugzilla itself to do this. It's also generally not a smart thing to do if you don't know exactly what you're doing. If you understand SQL, though, you can use the mysql command line utility to manually insert, delete and modify table information. There are also more intuitive GUI clients available. Personal favorites of the Bugzilla team are phpMyAdmin and MySQL Control Center.

Remember, backups are your friend. Everyone makes mistakes, and it's nice to have a safety net in case you mess something up. Consider using mysqldump to make a duplicate of your database before altering it manually.

Try running MySQL from its binary: mysqld --skip-grant-tables. This will allow you to completely rule out grant tables as the cause of your frustration. If this Bugzilla is able to connect at this point then you need to check that you have granted proper permission to the user password combo defined in localconfig.

Running MySQL with this command line option is very insecure and should only be done when not connected to the external network as a troubleshooting step.

You may also be suffering from a client version mismatch:

MySQL 4.1 and up uses an authentication protocol based on a password hashing algorithm that is incompatible with that used by older clients. If you upgrade the server to 4.1, attempts to connect to it with an older client may fail with the following message:

shell> mysql Client does not support authentication protocol requested by server; consider upgrading MySQL client

To solve this problem, you should use one of the following approaches:

• Upgrade all client programs to use a 4.1.1 or newer client library.

• When connecting to the server with a pre-4.1 client program, use an account that still has a pre-4.1-style password.

• Reset the password to pre-4.1 style for each user that needs to use a pre-4.1 client program. This can be done using the SET PASSWORD statement and the OLD_PASSWORD() function:

  mysql> SET PASSWORD FOR -> ' some_user '@' some_host ' = OLD_PASSWORD(' newpwd ');

Well, you can synchronize or you can move bugs. Synchronization will only work one way -- you can create a read-only copy of the database at one site, and have it regularly updated at intervals from the main database.

MySQL has some synchronization features built-in to the latest releases. It would be great if someone looked into the possibilities there and provided a report to the newsgroup on how to effectively synchronize two Bugzilla installations.

If you simply need to transfer bugs from one Bugzilla to another, checkout the "move.pl" script in the Bugzilla distribution.

Remove Windows. Install Linux. Install Bugzilla. The boss will never know the difference. B^)

Seriously though, making Bugzilla work easily with Windows was one of the major goals of the 2.18 milestone. If the necessary components are in place (perl, a webserver, an MTA, etc.) then installation of Bugzilla on a Windows box should be no more difficult than on any other platform. As with any installation, we recommend that you carefully and completely follow the installation instructions in Section 2.5.1.

While doing so, don't forget to check out the very excellent guide to Installing Bugzilla on Microsoft Windows written by Byron Jones. Thanks, Byron!

Not currently. Bundle::Bugzilla enormously simplifies Bugzilla installation on UNIX systems. If someone can volunteer to create a suitable PPM bundle for Win32, it would be appreciated.

Depending on what Web server you are using, you will have to configure the Web server to treat *.cgi files as CGI scripts. In IIS, you do this by adding *.cgi to the App Mappings with the <path>\perl.exe %s %s as the executable.

Microsoft has some advice on this matter, as well:

"Set application mappings. In the ISM, map the extension for the script file(s) to the executable for the script interpreter. For example, you might map the extension .py to Python.exe, the executable for the Python script interpreter. Note For the ActiveState Perl script interpreter, the extension '.pl' is associated with PerlIS.dll by default. If you want to change the association of .pl to perl.exe, you need to change the application mapping. In the mapping, you must add two percent (%) characters to the end of the pathname for perl.exe, as shown in this example: c:\perl\bin\perl.exe %s %s"

Your modules may be outdated or inaccurate. Try:

1. Hitting http://www.activestate.com/ActivePerl

2. Download ActivePerl

3. Go to your prompt

4. Type 'ppm'

5. PPM> install DBI DBD-mysql GD

New in 2.16 - go to the Account section of the Preferences. You will be emailed at both addresses for confirmation.

The interface was simplified by a UI designer for 2.16. Further suggestions for improvement are welcome, but we won't sacrifice power for simplicity.

As of 2.18, there is also a 'simpler' search available. At the top of the search page are two links; "Advanced Search" will take you to the familiar full-power/full-complexity search page. The "Find a Specific Bug" link will take you to a much-simplified page where you can pick a product and status (open,closed, or both), then enter words that appear in the bug you want to find. This search will scour the 'Summary' and 'Comment' fields, and return a list of bugs sorted so that the bugs with the most hits/matches are nearer to the top.

Matches in the Summary will 'trump' matches in comments, and bugs with summary-matches will be placed higher in the buglist -- even if a lower-ranked bug has more matches in the comments section.

Bugzilla uses a cookie to remember which version of the page you visited last, and brings that page up when you next do a search. The default page for new users (or after an upgrade) is the 'simple' search.

The current behavior is acceptable to bugzilla.mozilla.org and most users. If you want to change this behavior, though, you have your choice of patches:

The most likely cause is a very old browser or a browser that is incompatible with file upload via POST. Download the latest version of your favourite browser to handle uploads correctly.

In the Bugzilla administrator UI, edit the keyword and it will let you replace the old keyword name with a new one. This will cause a problem with the keyword cache; run sanitycheck.cgi to fix it.

Simple answer; you can.

The logic behind the page checks every bug in the list to determine legal state changes, and then only shows you controls to do things that could apply to every bug on the list. The reason for this is that if you try to do something illegal to a bug, the whole process will grind to a halt, and all changes after the failed one will also fail. Since that isn't a good outcome, the page doesn't even present you with the option.

In practical terms, that means that in order to mark multiple bugs as CLOSED, then every bug on the page has to be either RESOLVED or VERIFIED already; if this is not the case, then the option to close the bugs will not appear on the page.

The rationale is that if you pick one of the bugs that's not VERIFIED and try to CLOSE it, the bug change will fail miserably (thus killing any changes in the list after it while doing the bulk change) so it doesn't even give you the choice.

Gerv and Myk suggest a 2-space indent, with embedded code sections on their own line, in line with outer tags. Like this:

<fred>
[% IF foo %]
  <bar>
  [% FOREACH x = barney %]
    <tr>
      <td>
        [% x %]
      </td>
    <tr>
  [% END %]
[% END %]
</fred>

Myk also recommends you turn on PRE_CHOMP in the template initialization to prevent bloating of HTML with unnecessary whitespace.

Please note that many have differing opinions on this subject, and the existing templates in Bugzilla espouse both this and a 4-space style. Either is acceptable; the above is preferred.

Try this link to view current bugs or requests for enhancement for Bugzilla.

You can view bugs marked for 3.2 release here. This list includes bugs for the 3.2 release that have already been fixed and checked into CVS. Please consult the Bugzilla Project Page for details on how to check current sources out of CVS so you can have these bug fixes early!

This is well-documented in bug 49862. Ultimately, it's as easy as adding the "---" priority field to your localconfig file in the appropriate area, re-running checksetup.pl, and then changing the default priority in your browser using editparams.cgi.

1. Enter a bug into bugzilla.mozilla.org for the "Bugzilla" product.

2. Upload your patch as a unified diff (having used "diff -u" against the current sources checked out of CVS), or new source file by clicking "Create a new attachment" link on the bug page you've just created, and include any descriptions of database changes you may make, into the bug ID you submitted in step #1. Be sure and click the "Patch" checkbox to indicate the text you are sending is a patch!

3. Announce your patch and the associated URL (https://bugzilla.mozilla.org/show_bug.cgi?id=XXXXXX) for discussion in the newsgroup (mozilla.support.bugzilla). You'll get a really good, fairly immediate reaction to the implications of your patch, which will also give us an idea how well-received the change would be.

4. If it passes muster with minimal modification, the person to whom the bug is assigned in Bugzilla is responsible for seeing the patch is checked into CVS.

5. Bask in the glory of the fact that you helped write the most successful open-source bug-tracking software on the planet :)