Issue #3464
closed
Unable to sync multiple releases in single deb repo
Description
In order to keep the same relative-url to match upstream mirrors, trying to sync the releases: xenial, xenial-updates, xenial-security
CLI: pulp-admin deb repo create --repo-id 'xenial' --relative-url '/ubuntu/' --feed 'http://mirrors.us.kernel.org/ubuntu/' --releases 'xenial,xenial-updates,xenial-security' --components 'main' --architectures 'amd64'
This will allow a mirror like: http://mirrors.us.kernel.org/ubuntu/dists/ (see xenial*)
CLI: pulp-admin deb repo sync run --repo-id 'xenial'
An exception then occurs: "Exception: Checksum did not match"
Files
I found part of the issue is with Ubuntu's release file has Suite is the sub-dist folder and Codename as the overall dist.
For example:
xenial-security = https://mirrors.edge.kernel.org/ubuntu/dists/xenial-security/Release
xenial-updates = https://mirrors.edge.kernel.org/ubuntu/dists/xenial-updates/Release
xenial = https://mirrors.edge.kernel.org/ubuntu/dists/xenial/Release
The following at least corrects the relative-url path. It doesn't fix the checksum issue as I think importers/sync.py is using the release.codename to compare vs release.suite. (Haven't had a change to confirm yet)
/usr/lib/python2.7/site-packages/pulp_deb/plugins/distributors/distributor.py
<
repometa = aptrepo.AptRepoMeta(
codename=codename,
components=[comp.name for comp in rel_components],
architectures=list(architectures),
)
>
repometa = aptrepo.AptRepoMeta(
codename=release_unit.suite,
components=[comp.name for comp in rel_components],
architectures=list(architectures),
)
Would it be possible to use the release tag as the repo vs relaying on the "Release" file inside the repo?
I am not sure what you mean by "release tag". Can you please elaborate?
Hi, sure thing. If a repo is created like this:
pulp-admin deb repo create --repo-id 'update-channels' --relative-url '/ubuntu/' --feed 'http://mirrors.us.kernel.org/ubuntu/' --components 'main,universe,multiverse,restricted' --architectures 'amd64' --releases "xenial-updates,trusty-updates"
Could the distributor publish with "/ubuntu/dists/xenial-update" & "/ubuntu/dists/trusty-update" which is what the user provides as the release name. Today, it's using "/ubuntu/dists/" + codename
Any thoughts on getting this fixed? Thanks
I have been working on the relevant code for unrelated reasons so I feel qualified to add my two cents.
The problem I see, is that there is no clean way of fixing this problem that does not risk creating new problems for other cases. Using the user supplied release name in the `MetadataStep` of `distributor.py` would currently make this step dependent on information stored in the `repo_importer` data base collection. This is not desirable since distributors and importers are meant to exist independently of one another.
A better solution might be to add an additional `distribution` field to the `units_deb_release` collection to store the needed information where it belongs. This would require a data base change and by extension a data base migration. Not a trivial fix.
For now the workaround is to sync these ubuntu releases in separate repositories.
- Related to Issue #4138: Handling publish paths correctly added
- Status changed from NEW to CLOSED - WONTFIX
- Status changed from CLOSED - WONTFIX to NEW
Was closed as part of Pulp2 mass-fix, but reopening so the Debian plugin devs can determine what they want to do.
I initially expected this PR (https://github.com/pulp/pulp_deb/pull/83) to solve this issue.
However, it appears that this is actually an issue within "python-debpkgr".
I suspect debpkgr is tripping up for a similar reason. (That is, it is using "codename" as an internal dict key, even though codename is not always unique within a single repository.)
My PR fixes this kind of issue within pulp_deb but not within debpkgr.
Until someone volunteers a fix for python-debpkgr, the only solution/work around is to keep the various "xenial" releases (i.e "xenial", "xenial-updates", "xenial-security", etc.) in separate repositories.
- Triaged changed from No to Yes
I read this article! I hope you will continue to have such articles to share with everyone! thank you!
- Status changed from NEW to CLOSED - WONTFIX
There will be no additional work to fix this for Pulp 2.
It should work in Pulp 3.
If you find a similar issue in Pulp 3, feel free to open a new ticket.
Also available in: Atom
PDF
Add the distribution field to the DB models
Within a standard Debian repository structure, the term "distribution" refers to the unique string given by the path segment between the "dists/" folder, and some "Release" file (without the trailing slash).
Since each "Release" file in the directory structure is associated with exactly one unique distribution string, the terms "distribution" and "release" can be (and often are) used interchangably.
The distribution string is most commonly (but not always) given by either the "codename" or the "suite". The pulp_deb implementation prior to this commit, has assumed that the distribution string is always equal to codename, and has therefore imposed a uniqueness constraint on the codename for all releases/distributions within a single repository.
Since upstream repository sources make no such assumption and are not necessarily structured using the codename, this has lead to a plathora of unpredictable and buggy behaviour when synchronizing upstream repositories with 'codename != distribution'.
This change fixes these problems by introducing and using a "distribution" field for both the units_deb_release and units_deb_component collections.
revealed #4871 (depends on the fix for this issue) https://pulp.plan.io/issues/4871
ref #3464, #4055 https://pulp.plan.io/issues/3464 https://pulp.plan.io/issues/4055
fixes #4138, #4705, #4707 https://pulp.plan.io/issues/4138 https://pulp.plan.io/issues/4705 https://pulp.plan.io/issues/4707