MASTER_SITES points to yt-dl.org which has both A and AAAA RRs: $ host yt-dl.org yt-dl.org has address 95.143.172.170 yt-dl.org has IPv6 address 2001:1a50:11:0:5f:8f:acaa:177 Dual-stack nodes would fallback to 95.143.172.170 if AAAA failed, but IPv6-only nodes (with DNS64 or not) can only try AAAA, which is unreachable from at least AS8359 and AS207996: (AS8359) $ traceroute6 2001:1a50:11:0:5f:8f:acaa:177 traceroute6 to 2001:1a50:11:0:5f:8f:acaa:177 (2001:1a50:11:0:5f:8f:acaa:177) from 2001:db8::1, 64 hops max, 28 byte packets 1 2001:1a50:11:60::1 2109.865 ms !A 2999.323 ms !A 3199.868 ms !A (AS207996) $ traceroute6 2001:1a50:11:0:5f:8f:acaa:177 traceroute6 to 2001:1a50:11:0:5f:8f:acaa:177 (2001:1a50:11:0:5f:8f:acaa:177) from 2001:db8::2, 64 hops max, 28 byte packets 1 2a0a-e5c0-2-12--5.loves.ipv6.at.ungleich.ch 0.553 ms 0.613 ms 0.643 ms 2 2a01:330:1000:1003::1 8.677 ms 8.489 ms 8.561 ms 3 2a01:330:0:102::2 8.884 ms 8.677 ms * 4 10gigabitethernet1-4.core1.zrh1.he.net 9.361 ms 9.272 ms 9.185 ms 5 100ge8-1.core1.zrh2.he.net 5.551 ms 5.945 ms 5.870 ms 6 100ge15-1.core1.fra1.he.net 11.129 ms 10.995 ms 10.997 ms 7 decix-100g.e3.mk.de 11.820 ms 12.161 ms 12.027 ms 8 ae1.lr1.mk.de 12.694 ms ae1.lr2.mk.de 12.122 ms ae1.lr1.mk.de 12.580 ms 9 ae3.cr2b.mk.de 12.784 ms 11.690 ms 11.827 ms 10 2001:1640::4b:2 11.703 ms 11.686 ms 11.811 ms 11 2001:1a50:ffff:fffe::1042 13.008 ms 12.563 ms 12.413 ms 12 2001:1a50:11:60::1 12.009 ms 12.065 ms 11.998 ms 13 2001:1a50:11:60::1 3082.651 ms !A 2659.112 ms !A 3071.638 ms !A Upstream was notified via e-mail to hostmaster from SOA of zone in April, no response and no issue resolution since then. I think switching to USE_GITHUB (DNS64) or mirroring distfile on FreeBSD cluster would fix this, until the issue is fixed by yt-dl.org hostmaster.
Created attachment 225989 [details] patch Patch attached with commit message ("git am"). I used upstream provided tarball instead, as in https://cgit.freebsd.org/ports/commit/?id=29e87cd021e8fb0ef77948bc5e049f1283e2a493
Stick do DISTVERSION everywhere rather than mixing PORT and DISTVERSION and why not keep main site as fallback?
Created attachment 226032 [details] v2 > why not keep main site as fallback? GH likely has better bandwidth and uptime; I don't see the need to keep yt-dl.org. Hostmaster ignores mail, so if the website is down, it would likely remain down longer than any possible GH outage. I think it would not hurt to have yt-dl.org as a fallback, attached v2 after your review. A better solution would be to mirror the distfile at distcache.FreeBSD.org, which has AAAA RR without the need of DNS64/NAT64. I think only a committer can do that. > Stick do DISTVERSION everywhere rather than mixing PORT and DISTVERSION Right, I wanted to minimize the amount of change, so a maintainer is more likely to approve the patch. V2 changes more as you suggested. Thanks for review.
yt-dl.org link stil uses PORTVERSION, as a workaround for those who don't want to fetch it manually for now at least might be www/yt-dlp which is a compatible fork (uses github.com)
(In reply to Daniel Engberg from comment #4) Well, seems like I wasn't paying much attention and just copied it as-is... I was distracted with something else yesterday; this is trivial enough to change if someone commits it, sorry for that. > as a workaround for those who don't want to fetch it manually /etc/hosts entry (uses nat64.net): 2a00:1098:2b::95.143.172.170 yt-dl.org
Maintainer timeout.
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=1138b7cb7a81f7a0348788a7aca8ccc85aaf1367 commit 1138b7cb7a81f7a0348788a7aca8ccc85aaf1367 Author: Evgeniy Khramtsov <evgeniy@khramtsov.org> AuthorDate: 2021-07-25 10:44:21 +0000 Commit: Gleb Popov <arrowd@FreeBSD.org> CommitDate: 2021-07-25 10:46:09 +0000 www/youtube_dl: Add Github to MASTER_SITES. The original site is unreachable from IPv6-only networks. While there, switch to using DISTVERSION. PR: 256432 Tested by: arrowd Approved by: multimedia (maintainer timeout) www/youtube_dl/Makefile | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
Pushed in, thanks.