vcswatch reports that
this package seems to have a new changelog entry (version
13.19~, distribution
unstable) and new commits
in its VCS. You should consider whether it's time to make
an upload.
Here are the relevant commit messages:
commit fb3f2abdc40fb81a0259f10ab361519a94abf49c
Merge: 09defa8b 1f518d49
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Tue Dec 16 17:02:07 2025 +0000
Merge branch 'oracled-say_log' into 'main'
tag2upload-oracled: Introduce warn_log for more consistent warnings
See merge request dgit-team/dgit!355
commit 1f518d49dad8d01887233b4d0fb4caa86422cd80
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Tue Dec 16 16:21:46 2025 +0000
tag2upload-oracled: Introduce warn_log for more consistent warnings
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
commit 09defa8ba60eb7902c3533fcef4556a5ad9c8a41
Merge: 11e63d2c f8ef4abd
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Tue Dec 16 15:13:04 2025 +0000
Merge branch 'oracled-say_log' into 'main'
tag2upload-oracled logging tidying
See merge request dgit-team/dgit!354
commit f8ef4abd57238289dc3ed69fd66a064faec6eb8e
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Tue Dec 16 14:28:48 2025 +0000
tag2upload-oracled: say_log: Update comment
Outputting to systemd's journal is now settled practice for this service.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
commit d9e21bacf8218d1ff48fb763826c8f6092c66958
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Tue Dec 16 14:27:47 2025 +0000
tag2upload-oracled: Replace calls to warn() with say_log()
Both output to STDERR but warn() does so in a less informative way.
This also allows us to change where warnings go more easily.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
commit c777ddfa223df85ae78126bd5f60562a89bff490
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Tue Dec 16 14:26:35 2025 +0000
tag2upload-oracled: Replace another die() with fail(), add a comment
See commit 7d7613376c51bcc9600f83e38512f21141ea1268.
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
commit 11e63d2c556ca72fc4fd5388b930758a1c9a1a8b
Merge: 46fd97fd 43b0d328
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Mon Dec 15 21:44:46 2025 +0000
Merge branch 'test-race' into 'main'
tests: Fix a protocol violation (race) by fake t2usm
See merge request dgit-team/dgit!353
commit 43b0d32850555b8371788c13dd58b1a564d9bef2
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Sun Dec 14 21:52:40 2025 +0000
tests: Fix a protocol violation (race) by fake t2usm
What is being echoed here is the human-directed MESSAGE. It was mistakenly
sent to stdout (the oracled connection) rather than stderr (for debugging).
This is a gross protocol violation since MESSAGE isn't a valid
manager-to-oracle message. But we don't normally notice because at this point
two things are happening:
- oracled is tearing down what it thinks of as the builder VM
(autopkgtest-virt-*0 and gearing up for a new job.
It does this before reading the next message.
- the "response script" reads the rest of the response and exits
and t2u-fake-manager sends oracled a SIGTERM.
Empirically, almost always, the SIGTERM arrives before oracled reads the
improper message. We recently saw one case in ci.d.n where it didn't.
I can repro the failure every time by adding `sleep 1` to t2u-fake-manager
just before the call to kill. And then, this commit fixes. it.
IMO this kind of bug is inevitable given the complex structure of the tests.
I'll file a bug about that.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit df3570718ff03f4ad60da1a322c436b4a9aac9f8
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Sun Dec 14 21:52:37 2025 +0000
tests: Fix a protocol violation (race) by fake t2usm (pre-fmt)
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 46fd97fd64db2b2d2c27082dd9793512e6d4035f
Merge: fea26227 e51603de
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Sat Dec 13 11:31:08 2025 +0000
Merge branch 'retry-protocol' into 'main'
t2u o2m protocol: Proposed retries
See merge request dgit-team/dgit!319
commit e51603defe798c2c629806b27eec23c1a5cb5f37
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Sat Dec 13 00:46:44 2025 +0000
t2u o2m protocol: say `commit-to-public-upload` is protocol 6 only
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 84fe527bf6f4cbfa59d86ae1d82e83f53948ea07
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Sat Dec 13 00:45:14 2025 +0000
t2u o2m protocol: Make RETRY-STATUS into an argument to job.
As Sean writes in
https://salsa.debian.org/dgit-team/dgit/-/merge_requests/319#note_703578
The Oracle should have an ID for the job as soon as possible for logging,
and also having the ID right away improves the general intelligibility of
the protocol to humans.
So just make the retry status part of the job line.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 08808dc933f6f9e16881bb3331d71bbfcb2cc2ce
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Fri Dec 12 20:36:17 2025 +0000
t2u o2m protocol: Make job-retry into one insn with an argument
This makes the processing and the description more regular.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 20a3a25999d6b1b8e219db7060029291f97dbc17
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Fri Dec 12 20:34:26 2025 +0000
t2u o2m protocol: Provide the retry info first
The `data-block` part is logically part of the `job` - kind of an attachment
to it. Let's not make that an attachment to retry-*.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit db6799ea13a3755c5742c4ed7235b119552da9b1
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Fri Dec 12 16:31:12 2025 +0000
t2u o2m protocol: Explicitly report whether email sent
We have some Oracle-side failures where the Oracle doesn't report - notably,
key not in keyring or signature problems.
We want an email report from the Manager for this.
This means the email report status needs to be explicitly in the protocol.
It was already weirdly implicit for retriable failures.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit e416e2e1a32dc61d2b143041c1fb27bfab27c7e4
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Fri Dec 12 16:28:04 2025 +0000
t2u o2m protocol: Indent the retriable error messages
I seem to have overlooked indenting this.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 13044d159cf92c5b5fb876f680b874cc42b85256
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Fri Dec 12 16:25:30 2025 +0000
t2u o2m protocol: Assign protocol version 6 to this version.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit ba1844def609c161ced7703e683a3e19e30a581d
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Thu Oct 2 21:09:32 2025 +0100
t2u: o2m protocol: Always send a message about the attempt's "lastness"
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit fd8406329c3fa15a83ff729bd928895a6874d10e
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Wed Sep 24 14:00:57 2025 +0100
t2u o2m protocol: Be even more explicit about the Oracle and retries
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 128c2c919156eb363fe8d5575d9125b8dca7b96e
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Thu Sep 18 18:41:06 2025 +0100
t2u o2m protocol: clarify that *all* policy is Manager's job
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 6086895895f7b3d09a8490b49f97537dff396ddc
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Wed Sep 17 16:54:54 2025 +0100
t2u o2m protocol: clarify that policy is Manager's job
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 40ec5b536f79a5b77e448813fd9efcf10d4082c8
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Wed Sep 17 16:53:17 2025 +0100
t2u o2m protocol: rename commitment point message
As per
https://salsa.debian.org/dgit-team/dgit/-/merge_requests/319#note_654055
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 738ca5a057b280bf0b65a439312b4e2ce28296a2
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Tue Sep 2 20:30:43 2025 +0100
t2u o2m protocol: Proposed retry policy
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 6156c9b0767101d4a07c012ffc92b8533f8fd0ba
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Tue Sep 2 20:31:48 2025 +0100
t2u o2m protocol: Provide for retries
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit 84a93e961ff5c3444307116210fd003e90a637a4
Author: Ian Jackson <ijackson@chiark.greenend.org.uk>
Date: Tue Sep 2 19:50:35 2025 +0100
t2u o2m protocol: Indent all the protocol messages
I'm going to want to annotate one with "opt" for optional.
Splitting this out will make the next commit clearer.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
commit fea262273d388b157142b2f9548f584a9744b3a0
Author: Sean Whitton <spwhitton@spwhitton.name>
Date: Fri Dec 12 15:50:41 2025 +0000
changelog: Start 13.19
Signed-off-by: Sean Whitton <spwhitton@spwhitton.name>
![[bug history graph]](https://qa.debian.org/data/bts/graphs/d/dgit.png){kind=link}