I will note that back around 2025-05-24 I did an experimental build sequence starting with 'bulk -Ca' as part of testing pkg 2.1.* . For that, all of the mongodb[5678]0* built just fine:
# ls -C1 /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_*/logs/mongodb[0-9]* /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_11h51m15s/logs/mongodb50-5.0.31_1.log /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_11h51m15s/logs/mongodb60-6.0.23_1.log /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_23h41m42s/logs/mongodb70-7.0.20.log /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_23h41m42s/logs/mongodb70-armv80a-7.0.20.log /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_23h41m42s/logs/mongodb80-8.0.9.log /usr/local/poudriere/data/logs/bulk/release-aarch64-default/2025-05-24_23h41m42s/logs/mongodb80-armv80a-8.0.9.log # uname -apKU FreeBSD aarch64-main-pbase 15.0-CURRENT FreeBSD 15.0-CURRENT main-n277335-7fa19ee28c90 GENERIC-NODEBUG arm64 aarch64 1500043 1500043 (That is an official PkgBase distribution.) # ~/fbsd-based-on-what-commit.sh -C /usr/ports/ aea1b30840dd (HEAD -> main, freebsd/main, freebsd/HEAD) */*: Remove version specifiers on go ports I maintain Author: Nuno Teixeira <[email protected]> Commit: Nuno Teixeira <[email protected]> CommitDate: 2025-05-22 02:08:34 +0000 branch: main merge-base: aea1b30840ddafa7cf18406c7454e72334b14be1 merge-base: CommitDate: 2025-05-22 02:08:34 +0000 n706807 (--first-parent --count for merge-base) As the package builder configurations are not public as far as I know, there was no effort to configure any sort of matching context to the official builders' configuration details. (I do expect that the the official builders all likely use ZFS, not UFS.) It was an aarch64 context with a UFS file system and USE_TMPFS=all but with an extensive TMPFS_BLACKLIST assignment. The number of builders was the number of FreeBSD CPUs, which in turn was the number of CPUs assigned to the Parallels Virtual machine, which was the number performance cores on the macOS system: 12. ALLOW_MAKE_JOBS=yes was in use without use of MAKE_JOBS_NUMBER_LIMIT or the like to constrain the load averages.(So, a high load average style of configuration.) RAM+SWAP totaled to 300 GiBytes, RAM being 64 GiBytes (assigned to the Parallels Virtual Machine). The storage media was a U.2 Optane via a USB3.2 capable adaptor. That media can boot and operate any of the aarch64 systems that I have access to (given an EDK2 ACPI context or an appropriate U-Boot UEFI/Device-Tree context is also present) --other than the Apple system. (It was not a VM file on macOS.) === Mark Millard marklmi at yahoo.com
