Bug 2135215

Attaching an issue for allowing the disablement of .treeinfo handling.  Faced with misbehaving webservers that lie about having a .treeinfo file, this is probably the best workaround.

In this specific case I'm still waiting to here back from the vendor about fixing their configuration.

https://github.com/pulp/pulp_rpm/issues/2848

Ian, what do you think about having an option available in the UI which controls whether or not kickstart metadata is synced?  How is this handled for SRPMs, is it also a checkbox or is it a manually-typed list of skippable types?

Daniel, the SRPM ignoring is done in the UI via a checkbox. We could add another checkbox for ignoring kickstart media. Will users know to ignore kickstart media when they see this treeinfo error?

(In reply to Ian Ballou from comment #9)
> Daniel, the SRPM ignoring is done in the UI via a checkbox. We could add
> another checkbox for ignoring kickstart media. Will users know to ignore
> kickstart media when they see this treeinfo error?

The short answer is "possibly/probably no", because the particular use-case that generated the error here arises from a misconfigured upstream-remote and shouldn't happen often.

There are four use-cases I can imagine for a user wanting to check a "skip kickstart-trees" box:

* You don't use kickstarts, and want to save time/space on sync and storage. User would know what the box means.

* You use kickstart trees, but your remote has a bad file/image which throws an obvious "checksum mismatch" error on failed syncs. User would probably be able to know to a) check this box and retry while b) yelling at their upstream about their content being broken. (This is the orginating error for pulp issue #2848)

* You use kickstart trees, but your remote has a malformed **treeinfo** file. You get the error listed here, and then do a) and b) above.

* The remote you're syncing from **doesn't have a kickstart treeinfo**, but cheerfully lies about it, returning a "200 found" while handing back a random piece of HTML saying "Not Found". This is the use-case that resulted in this BZ. User may or may not have any idea what's going on - the remote doesn't have a kickstart tree, why would I check this box? This kind of misconfiguration is rare.

I would give such a box a "break in case of..." description associated with it.

The Pulp upstream bug status is at closed. Updating the external tracker on this bug.

All upstream Pulp bugs are at MODIFIED+. Moving this bug to POST.

Ian, is there a BZ filed for a checkbox which allows omitting treeinfo metadata from the sync?

(re: comment #9)

BZ for Satellite side: https://bugzilla.redhat.com/show_bug.cgi?id=2174912

*** Bug 2176870 has been marked as a duplicate of this bug. ***

This BZ and https://bugzilla.redhat.com/show_bug.cgi?id=2174912 can be verified at the same time by using this new mechanism.

Verified in 6.14.0 snap 4.0.

Steps to verify:
1) Create a product and repo with upstream url from comment#2, try to sync it (try optimized, complete).
2) Set 'Ignore treeinfo' of that repository to 'Yes', try to sync it.
3) Retry with different mirroring (all but 'Complete Mirroring') and download policies.

Expected / observed results:
1) A decent descriptive error message was shown:
   "This repository has a malformed or inaccessible treeinfo file. To sync, try enabling 'Ignore treeinfo'. All kickstart contents will be skipped."
2) Sync succeeded and the RPM content was available at the Published_at url.
3) Sync succeeded for all allowed combinations and proper validation error was shown when attempted to set 'Treeinfo ignore' && 'Complete mirroring'

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Important: Satellite 6.14 security and bug fix update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHSA-2023:6818