size mismatch error if request of unknown size is larger than others

Please let me know if you can reproduce this with 2.2.2 in hirsute. No changes can be made to the http method in stable releases, as it's super fragile - Every time we make changes, something else breaks.

Managing to get a reproducer for HTTP issues more than a couple of hours has not really worked so far, and most people's reproducers don't work. It's highly subjective to network latency and MTU afaict.

> I've attached an example log, where the error pops up for multiple packages, and they all appear to be compared to one size (86464 bytes).

just for the record: This is a misunderstanding. If apt does pipelining it searches in the requests it made for the file this response is for. If no request matches the size the error shows the last tried size (aka of the last file in the queue).

I wonder a bit why Filesize isn't included in the message as internally it is implemented as a (very weak) hash, but that might be due to the apt version… I am not remembering ATM how it worked back then.

I am also a bit worried about the screenshot in the referred bugreport as that shows two different servers replying (Apache vs some python via a proxy).

And last not least: The sizes given in the HTTP request logs for python3-zmq (which failed in that example) shown in /var/log/httpd/foreman_access.log and /var/log/messages do not match (254232 vs 254454).

> I am also a bit worried about the screenshot in the referred bugreport as that shows two different servers replying (Apache vs some python via a proxy).

The saltstack repository is served by pulp, so client -> httpd -> pulp. I've verified that httpd responds with correctly. Not sure why pulpcore response is slightly larger, maybe something to do with wsgi.

> Please let me know if you can reproduce this with 2.2.2 in hirsute

I don't have that mirrored, but I have focal, and I could not reproduce. Apt 2.0.2

> I don't have that mirrored, but I have focal, and I could not reproduce. Apt 2.0.2

There have been no relevant changes between 1.6.y and 2.0.y, so it really should reproduce with both of them or none of them.

But you have to be careful that the download is _exactly_ the same, usually by pointing new apt at the older system state, otherwise you might have lost your reproducer.

But yeah, this "comes and goes" is why we're not making much progress on HTTP bugs.

With help from @juliank, I was able to transfer the state from my bionic box to focal, and hirsute. Issue reproduces on all of them. On hirsute, apt is 2.2.1. Where do we go from here?

We tracked this down to the Pulp repository not having Size fields. While this is a broken repository, APT should also more gracefully handle that situation, either by warning about missing Size values, erroring out, disabling pipelining; or by fixing pipelining to behave if we don't know the size in advance.

I tracked it down further today, accidentally finding a reproducer. In methods/http.cc we check for Req.DownloadSize > Req.MaximumSize and fail with an error if that's the case; hence ensuring our download was within the allowed range for the request.

The issue is that Req.MaximumSize is calculated by BaseHttpMethod::FindMaximumObjectSizeInQueue as the maximum of all maximum sizes in the queue; and requests with unknown sizes are encoded as 0, so FindMaximumObjectSizeInQueue() will return any other size that is not 0, despite the 0 meaning "unknown". The fix is to return 0 if any maximum size is 0.

This bug was fixed in the package apt - 2.2.3

---------------
apt (2.2.3) unstable; urgency=medium

  * tests: Check for and discard expected warning from MaybeAddAuth. For some
    reason, this was only noticed with LTO enabled, but should be a general
    issue.
  * Fix downloads of unsized files that are largest in pipeline (LP: #1921626)
  * Warn on packages without a Size field. Such repositories are broken and
    need to be fixed, as we do not test apt against them, see the bug above
    for more details. Set Acquire::AllowUnsizedPackages to disable the
    warning.

 -- Julian Andres Klode <email address hidden> Tue, 13 Apr 2021 17:53:32 +0200

Will this fix be packported to bionic and focal?

Yes, see the tasks.

Hello Rossen, or anyone else affected,

Accepted apt into focal-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/apt/2.0.6 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-focal to verification-done-focal. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-focal. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

Hello Rossen, or anyone else affected,

Accepted apt into bionic-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/apt/1.6.14 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-bionic to verification-done-bionic. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-bionic. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

All autopkgtests for the newly accepted apt (1.6.14) for bionic have finished running.
The following regressions have been reported in tests triggered by the package:

apport/2.20.9-0ubuntu7.24 (amd64)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/bionic/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

All autopkgtests for the newly accepted apt (2.0.6) for focal have finished running.
The following regressions have been reported in tests triggered by the package:

reprotest/0.7.14 (ppc64el)
apport/2.20.11-0ubuntu27.18 (amd64)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/focal/update_excuses.html#apt

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

autopkgtests for apt itself have passed, verifying the fix. We still battle with one apport regression on focal, but it seems unrelated - they've failed before.

This bug was fixed in the package apt - 2.0.6

---------------
apt (2.0.6) focal; urgency=medium

  * RunScripts: Do not reset SIGQUIT and SIGINT to SIG_DFL (LP: #1898026)
  * test suite fix: Check for and discard expected warning from MaybeAddAuth
  * Fix downloads of unsized files that are largest in pipeline (LP: #1921626),
    and warn about packages without size (option Acquire::AllowUnsizedPackages)
  * JSON hooks 0.2 and assorted JSON bugfixes (LP: #1926150)
    - encoder fixes:
      + json: Escape strings using \u escape sequences, add test
      + json: Actually pop states
      + json: Encode NULL strings as null
    - json: Flush standard file descriptors before calling hooks
      (this avoids output from hooks in middle of apt output)
    - non-code changes:
      + test/json: Make the test hook more reliable
      + Fix a typo in json-hooks-protocol.md (thanks to Brian Murray)
    - semantic changes (new fields, hooks, and protocol 0.2):
      + json: Add origins fields to version
      + upgrade: Add JSON hook support (AptCli::Hooks::Upgrade)
      + json: Add `package-list` and `statistics` install hooks
      + json: Hook protocol 0.2 (added upgrade,downgrade,reinstall modes)
    + Fix a typo in json-hooks-protocol.md (thanks to Brian Murray)
  * Avoid infinite loop on EOF on media change prompt (LP: #1928687)

 -- Julian Andres Klode <email address hidden> Tue, 15 Jun 2021 13:05:46 +0200

The verification of the Stable Release Update for apt has completed successfully and the package is now being released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.

This bug was fixed in the package apt - 1.6.14

---------------
apt (1.6.14) bionic; urgency=medium

  * RunScripts: Do not reset SIGQUIT and SIGINT to SIG_DFL (LP: #1898026)
  * Fix downloads of unsized files that are largest in pipeline (LP: #1921626),
    and warn about packages without size (option Acquire::AllowUnsizedPackages)
  * JSON hooks 0.2 and assorted JSON bugfixes (LP: #1926150)
    - encoder fixes:
      + json: Escape strings using \u escape sequences, add test
      + json: Actually pop states
      + json: Encode NULL strings as null
    - json: Flush standard file descriptors before calling hooks
      (this avoids output from hooks in middle of apt output)
    - Minor fixes to include and C++ namespaces
    - non-code changes:
      + test/json: Make the test hook more reliable
      + Fix a typo in json-hooks-protocol.md (thanks to Brian Murray)
    - semantic changes (new fields, hooks, and protocol 0.2):
      + json: Add origins fields to version
      + upgrade: Add JSON hook support (AptCli::Hooks::Upgrade)
      + json: Add `package-list` and `statistics` install hooks
      + json: Hook protocol 0.2 (added upgrade,downgrade,reinstall modes)
    + Fix a typo in json-hooks-protocol.md (thanks to Brian Murray)
  * Avoid infinite loop on EOF on media change prompt (LP: #1928687)

 -- Julian Andres Klode <email address hidden> Tue, 15 Jun 2021 16:12:38 +0200