Some archives are empty; ./+CONTENTS not found in archive

[mseri (Marcello Seri)]

This is the output. I have recently upgraded to Big Sur. It happened when upgrading the outdated packages:

--->  Computing dependencies for llvm-9.0
--->  Fetching archive for llvm-9.0
--->  Attempting to fetch llvm-9.0-9.0.1_1.darwin_20.x86_64.tbz2 from https://packages.macports.org/llvm-9.0
--->  Attempting to fetch llvm-9.0-9.0.1_1.darwin_20.x86_64.tbz2.rmd160 from https://packages.macports.org/llvm-9.0
--->  Installing llvm-9.0 @9.0.1_1
tar: ./+CONTENTS: Not found in archive
tar: Error exit delayed from previous errors.
Error: Failed to install llvm-9.0: child process exited abnormally
Error: See /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_lang_llvm-9.0/llvm-9.0/main.log for details.
Error: Problem while installing llvm-9.0
Error: Follow https://guide.macports.org/#project.tickets to report a bug.

[mseri (Marcello Seri)]

upgrade error log

[ryandesign (Ryan Carsten Schmidt)]

The file llvm-9.0-9.0.1_1.darwin_20.x86_64.tbz2 on our server appears to be completely empty. I'm investigating.

[ryandesign (Ryan Carsten Schmidt)]

Has duplicate #61499 for rust.

The archives were generated correctly on the buildbot worker but something went wrong when uploading them to the master.

There are other affected files too. I will need to identify all the broken files on the master server and delete them and then schedule those builds on the workers again to reupload them.

The master server has an uptime of 132 days and I have in the past seen weird behavior on the server when it is online for that long.

I'll want to wait for the current builds to finish, then restart the server, and hopefully that lets it behave normally again.

[ryandesign (Ryan Carsten Schmidt)]

Has duplicate #61505 for go.

Bad archives deleted so far:

-rw-r--r--  1 buildbot  buildbot    6757975 Nov 16 14:18 bit/bit-0.9.11_0.darwin_20.x86_64.tbz2
-rw-r--r--  1 buildbot  buildbot        512 Nov 16 14:19 bit/bit-0.9.11_0.darwin_20.x86_64.tbz2.rmd160
-rw-r--r--  1 buildbot  buildbot  124528862 Nov 16 09:37 go/go-1.15.5_0.darwin_20.x86_64.tbz2
-rw-r--r--  1 buildbot  buildbot        512 Nov 16 09:45 go/go-1.15.5_0.darwin_20.x86_64.tbz2.rmd160
-rw-r--r--  1 buildbot  buildbot   48153614 Nov 16 14:07 llvm-9.0/llvm-9.0-9.0.1_1.darwin_20.x86_64.tbz2
-rw-r--r--  1 buildbot  buildbot        512 Nov 16 14:10 llvm-9.0/llvm-9.0-9.0.1_1.darwin_20.x86_64.tbz2.rmd160
-rw-r--r--  1 buildbot  buildbot          0 Nov 16 14:10 rust/rust-1.47.0_0.darwin_20.x86_64.tbz2
-rw-r--r--  1 buildbot  buildbot        512 Nov 16 14:10 rust/rust-1.47.0_0.darwin_20.x86_64.tbz2.rmd160

[ryandesign (Ryan Carsten Schmidt)]

The server also corrupted the git clone used for mirroring:

​https://build.macports.org/builders/jobs-mirror/builds/216185/steps/checkout/logs/stdio

fatal: .git/index: index file smaller than expected

I've deleted that clone so it should be recreated for the next build and mirroring should work correctly from then on.

[ryandesign (Ryan Carsten Schmidt)]

Has duplicate #61510 (llvm-9.0).

I've rebooted the server and am currently checking the disks with Disk Utility's First Aid function.

[ryandesign (Ryan Carsten Schmidt)]

A minor problem with free space on the boot volume was fixed. No problems were found on the RAID which stores all the data.

[ryandesign (Ryan Carsten Schmidt)]

I checked all private (nondistributable) archives and found only one that was identified as "data" (and it contained all zeroes) and have removed it:

-rw-r--r--  1 buildbot  buildbot  11643757 Nov 16 08:29 spring-boot-cli/spring-boot-cli-2.4.0_0.darwin_20.noarch.tbz2
-rw-r--r--  1 buildbot  buildbot       512 Nov 16 08:29 spring-boot-cli/spring-boot-cli-2.4.0_0.darwin_20.noarch.tbz2.rmd160

I also purged these 10 files from the CDN.

I found three private archives that were identified by file as "DOS/MBR boot sector" but I think that's just a quirk of file; they seem to be valid tbz2 files and to extract just fine. Those files are:

-rw-r--r--  1 buildbot  buildbot  107906829 Oct  7 02:18 mariadb-10.4/mariadb-10.4-10.4.14_1.darwin_19.x86_64.tbz2
-rw-r--r--  1 buildbot  buildbot     900847 Nov  7  2019 pstoedit/pstoedit-3.74_1.darwin_17.x86_64.tbz2
-rw-r--r--  1 buildbot  buildbot     407745 Jan 29  2020 universal-ctags/universal-ctags-20200111_0+iconv+libxml2+manpages.darwin_16.x86_64.tbz2

I'm still checking if there are any more affected public archives.

[ryandesign (Ryan Carsten Schmidt)]

No further affected public archives. Just two more valid tbz2 archives identified by file as "DOS/MBR boot sector":

-rw-r--r--  1 buildbot  buildbot      3880 Jun  4  2017 ./p5.26-data-uniqid/p5.26-data-uniqid-0.120.0_0.darwin_10.noarch.tbz2
-rw-r--r--  1 buildbot  buildbot  12720605 Sep 11  2019 ./py37-jupyterlab/py37-jupyterlab-1.1.3_0.darwin_10.noarch.tbz2