Apache Allura™

• milestone: forge-backlog --> someday

Originally by: nathancarter

I am having trouble finding on the internet anywhere (or SF.net in particular) a reliable way to take an existing Trac (hosted app) ticket database and import it into Allura. (E.g., the support discussion at https://sourceforge.net/p/forge/site-support/1701/ says that the official migration documentation is incorrect, and maybe the Allura devs have a better answer, but no such official answer is posted anywhere that I can find.) Despite this, if I understand correctly, it is required to update to Allura in the somewhat near future. So is the official position that those with existing Trac hosted app ticket databases are simply left out in the cold?

There are 2 separate things going on:

1. Projects should upgrade to Allura. Hosted Apps (e.g. Trac) will still work in Allura. So you can safely do this without losing anything.
2. Hosted Apps will be decommissioned at some point TBD. The recommended path for Trac is to install Trac in the "project web" space. That is explained at https://sourceforge.net/p/forge/community-docs/Migrating%20Trac%20from%20Hosted%20Apps/ but is incomplete so far. We'll publish more info on that page or on the SourceForge blog as more details about Hosted Apps transition are determined.

Someday we might provide a path from Trac into Allura, but there's no plan for that right now.

Originally by: nathancarter

That does help clarify, thank you!

In that same spirit, please allow me to explain my three options and let me know if I understand them correctly.

1. Upgrade to Allura without moving Trac anywhere: Trac continues to work for the short term (until Hosted Apps are decommissioned one day), but Trac-SVN integration (important to us) goes away.
2. Upgrade to Allura and then (after the page you linked to has all its known issues resolved) move Trac to project web space: Trac continues to work indefinitely, and I can even install new packages into Trac since now I control the installation, but Trac-SVN integration still goes away. It may come back, however, since the post you linked to says they're working on it.
3. Perform one of the above upgrades, and then move our source code to use git. Will Trac integrate with git in either of the above two scenarios, or is that lost, just like SVN integration is?

Thank you for helping us make an informed plan!

Yep, you've got it. Unfortunately there isn't SVN (or git) integration with Trac when Trac is installed on project web space. I don't know offhand if that will be made available or not.

Originally by: *anonymous

So the path forward seems to be this:

1. Upgrade to Allura. Don't move Trac.
2. When the instructions for moving Trac to project web space are working (including Trac-SVN integration), perform that transfer. (SF staff aim to ensure that this precedes Hosted App decommissioning.)
3. Add whatever Trac packages I want after that.

That's how I'll proceed. Thanks again!

• Description has changed:

Diff:

--- old
+++ new
@@ -1,3 +1,10 @@
-We should have a script to import from a trac database dump into a wiki tool and ticket tool.  This probably should be in 2 separate parts, or have options to run just the wiki or just the ticket parts, since some trac instances only use one or the other.
+A script exists at `scripts/trac_export.py` to extract trac tickets, and can be imported with `scripts/allura_import.py`  Jon Schewe describes the whole process at <http://mail-archives.apache.org/mod_mbox/incubator-allura-users/201303.mbox/%3CCALpcnoxEG%3DNJ7UG1dfOKj%2BbEPMtybGNStXZPVjfHrLyLDz%3DXLA%40mail.gmail.com%3E>

-This should be done after [#4186] is completely done.  The wiki part can share a lot with that.  In fact, the loading script from [#4186] should be completely re-usable.  We just need a trac extractor to create the data files in the proper format.
+Improvements needed:
+
+* stability / handle errors
+* fetch and upload attachments (add uploads to import API?)
+* set up milestones and proper field types (add to API? or setup directly, first?)
+* SourceForge: a task for us to run it for projects, including the user mapping
+
+See [#6139] for wiki import.

• Milestone: someday --> forge-backlog

• summary: Import from trac into wiki & ticket tools --> Import from trac into ticket tool
• Description has changed:

Diff:

--- old
+++ new
@@ -8,3 +8,5 @@
 * SourceForge: a task for us to run it for projects, including the user mapping

 See [#6139] for wiki import.
+
+We should strongly consider making our ticket import format match our output format, so that a backup [#3154] can be restored directly.

• Milestone: forge-backlog --> forge-may-03

• Description has changed:

Diff:

--- old
+++ new
@@ -9,4 +9,4 @@

 See [#6139] for wiki import.

-We should strongly consider making our ticket import format match our output format, so that a backup [#3154] can be restored directly.
+We should strongly consider making our ticket import format match our API format, so that a backup [#3154] can be restored directly. ... or do that later

• Size: --> 4

• Description has changed:

Diff:

--- old
+++ new
@@ -2,11 +2,11 @@

 Improvements needed:

-* stability / handle errors
+* stability / handle errors (and possibly some refactor/cleanup)
 * fetch and upload attachments (add uploads to import API?)
 * set up milestones and proper field types (add to API? or setup directly, first?)
-* SourceForge: a task for us to run it for projects, including the user mapping
+* SourceForge: a ScriptTask for us to run it for projects, including the user mapping

 See [#6139] for wiki import.

-We should strongly consider making our ticket import format match our API format, so that a backup [#3154] can be restored directly. ... or do that later
+<s>We should strongly consider making our ticket import format match our API format, so that a backup [#3154] can be restored directly.</s> ... or do that later

• Milestone: forge-may-03 --> forge-may-17

• labels: import --> import, 42cc

Created:

• #344: [#4213] Import from trac: fetch and upload attachments (3cp)
• #345: [#4213] Import from trac: milestones and field types (3cp)
• #346: [#4213] Import from trac: errors and refactoring (2cp)
• #347: [#4213] Import from trac: ScriptTask to run import (2cp)

• status: open --> in-progress

• Milestone: forge-may-17 --> forge-backlog

Closed #344. je/42cc_4213

Closed #345. je/42cc_4213

Originally by: nathancarter

I'm excited to see progress on this ticket! Thanks, Igor!
To be sure I understand what's happening: (1) We've already got a Trac export method described in the link in the description of this ticket. (2) Once you're done with #344-#347, we'll have an automated process for pulling that extracted stuff into Allura (YAY!). (3) Therefore we can be totally done with Trac and switch to Allura, which has the scm integration (both svn and git) that Trac wouldn't have if it were installed in project web space.
In other words, as soon as you do #346 and #347, we have a path to upgrade Trac projects to Allura ASAP. Have I understood this correctly? (If so, allow me to say yay once again!)

Yep, that's right.

In fact, most of this work is improvements to the process, and the basic process works already.

Originally by: nathancarter

Great, thank you!

Originally by: nathancarter

OK I've done the trac export and am ready to move on to allura_import.py. I'm to the point where I need an API key. But Googling around doesn't seem to tell me where I go about requesting one. Can you tell me how to start that process?

Hello Nathan, please log support ticket for this, make sure to indicate the project this is for, and mark the ticket as private. I'll generate a key for you and post the info there.

Originally by: nathancarter

Done. Thank you!

Originally by: nathancarter

Running allura_import.py in --validate mode completes, and the only warnings are about usernames, and it's just a handful. I plan to fix them manually after the real import.

So I proceed to do the real import (exact same command without the --validate) and I get a 403 Forbidden, with the only text of the page being, "We're sorry but we weren't able to process this request." Not sure what to do next because that's not very informative...

Any tips? Sorry to be a pest here.

Closed #346. je/42cc_4213