From: Piotr P. Karwasz <[email protected]> To: [email protected] Subject: ASF Security Team evaluating Allura as a security-report tracker MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit
Hi all, We're looking for a replacement for how the ASF Security Team tracks inbound vulnerability reports, and Allura is our leading candidate. Before we commit, we'd like your view on two things: whether the changes we'd need could be accepted upstream, and whether building on Allura is sustainable from your side. The problem Today we track reports sent to [email protected] with Gmail labels, plus a small in-house dashboard that gives each PMC access to its own reports. This no longer scales. We want a proper tracker: one private Allura instance, one neighbourhood, and one project for each of the ~200 PMCs. Each report becomes a ticket, and PMC members can see only their own project's tickets. What we found The tracker side fits well: custom fields, configurable statuses, labels, search, history and the REST API all do what we need. The entry-point architecture also means we shouldn't need a fork. We'd still need to build some things, and a few of them are generic enough that they may belong upstream: - Inbound mail routing. Allura routes incoming mail only by recipient address, so mail sent to a single fixed address such as [email protected] has nowhere to land. We'd write a gateway that matches follow-ups to tickets by Message-ID. A small, generic hook in the mail-handling path would make this much cleaner than re-injecting mail into the SMTP listener. - Group-to-role mapping. Allura's LDAP support handles authentication only. We'd write an ASF OAuth provider that syncs PMC membership into project roles at login. That provider is ASF-specific and would probably stay in our own repository as a plugin. A generic "sync external groups to roles" hook in the auth provider API might be useful upstream, though. - Read-access audit logging. We need a record of who viewed a restricted ticket, especially when the viewer isn't a member of the project. - Tracker statistics. We need a chart of open/closed reports over time for each project. Our questions 1. Would you be open to accepting changes like these upstream, and do you have a preference between core changes and new extension points? 2. Would you consider the extension points we'd rely on (Application, auth providers, template overrides, the mail handler) stable enough to build on? 3. How do you see the project's health over the next few years? Looking at the history, the project is clearly active, and the recent hardening work under #8624 was encouraging. Most commits come from a small group of people, though. Since the Foundation's whole security intake would depend on this system, we want to ask openly whether you think this is sustainable, rather than guess. To be upfront: we can't promise a steady stream of contributions. Our main work is elsewhere. We would upstream the things we build if you want them, report bugs, and report any security issues we find through the usual process. Anything not accepted upstream we'd maintain ourselves as plugins. Any thoughts are welcome, including "this isn't a good fit." Thanks, Piotr
