From Newsgroup: muc.lists.freebsd.stable
--aafbdc66-34b5-4160-9698-0d88a3170554-1
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
August 1, 2026 at 2:25 PM, "Mark Millard" <
marklmi@yahoo.com mailto:markl=
mi@yahoo.com?to=3D%22Mark%20Millard%22%20%3Cmarklmi%40yahoo.com%3E > wrot=
e:
=20
=20On 8/1/26 08:46, Nenhum_de_Nos wrote:
=20
=20
August 1, 2026 at 10:47 AM, "Karl Denninger" <karl@denninger.net
<mailto:karl@denninger.net?
to=3D%22Karl%20Denninger%22%20%3Ckarl%40denninger.net%3E>> wrote:
=20=20
The problem appears to be at this point:
=20=20
Cloning into '/usr/src'...
fatal: unable to access 'https://git.freebsd.org/src.git/ <https://g=
it.freebsd.org/src.git/>':
https://git.freebsd.org/src.git/%3E': Resolvi=
ng timed out after 300226 milliseconds
=20=20
I do not know why, however, because my "base" source git clone has
in its config the following:
=20=20
[remote "freebsd"]
=C2=A0 =C2=A0 =C2=A0 =C2=A0 url =3D https://git.freebsd.org/src.git =
<
https://
git.freebsd.org/src.git>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 fetch =3D +refs/heads/*:refs/remotes/fre=
ebsd/*
=20=20
And it works.
=20=20
I have several worktrees checked out against that since I build fo=
r
14-STABLE and 15-STABLE at this point for distribution to machines
running ARM64 and AMD64.
=20=20
My latest "git pull --ff-only" to so-update was this morning.
=20=20
On 8/1/2026 09:36, Nenhum_de_Nos wrote:
=20=20
Hi,
=20=20
I have been fighting this issue for 2 weeks now, I guess its time =
to ask here. The problem:
=20=20
root@arm_build_hd:/usr # rm -rf src/
root@arm_build_hd:/usr # /root/git/freebsd_config/freebsd_git/src_gi=
t_checkout_15.1R.sh
Cloning into '/usr/src'...
fatal: unable to access 'https://git.freebsd.org/src.git/ <https://g=
it.freebsd.org/src.git/>':
https://git.freebsd.org/src.git/%3E': Resolvi=
ng timed out after 300226 milliseconds
root@arm_build_hd:/usr # /root/git/freebsd_config/freebsd_git/src_gi=
t_checkout_15.1R.sh
Cloning into '/usr/src'...
remote: Enumerating objects: 4850942, done.
remote: Counting objects: 100% (384686/384686), done.
remote: Compressing objects: 100% (29598/29598), done.
error: RPC failed; curl 56 OpenSSL SSL_read: OpenSSL/3.5.6: error:0A=
000126:SSL routines::unexpected eof while reading, errno 0
error: 73 bytes of body are still expected
fetch-pack: unexpected disconnect while reading sideband packet
fatal: early EOF
fatal: fetch-pack: invalid index-pack output
root@arm_build_hd:/usr # /root/git/freebsd_config/freebsd_git/src_gi=
t_checkout_15.1R.sh
you have mail
Cloning into '/usr/src'...
fatal: unable to access 'https://git.freebsd.org/src.git/ <https://g=
it.freebsd.org/src.git/>':
https://git.freebsd.org/src.git/%3E': Resolvi=
ng timed out after 300040 milliseconds
root@arm_build_hd:/usr # cat /root/git/freebsd_config/freebsd_git/sr=
c_git_checkout_15.1R.sh
git clone --branch releng/15.1 https://git.freebsd.org/src.git <http=
s://git.freebsd.org/src.git>
https://git.freebsd.org/src.git%3E /usr/src=
=20
root@arm_build_hd:/usr #=20
=20=20
On two Orange Pi 5 Plus 16GB I have FreeBSD 15.1-R installed so I =
can build opnsense for arm64 (this is not relevant for the problem, excep=
t for one detail I will tell later). This box above had 14.3-R running fi=
ne, all build times were fine, git worked perfectly. Since I needed to up= grade to 15.1-R, git stopped working as it was expected (and presented ab= ove).
=20=20
I looked for DNS reasons, changed DNS server, local server, remote=
server, kept ping github.com while cloning, no good. I started using the=
upgraded box, then I got another similar arm64 box and installed fresh 1= 5.1-R, everytime the same issue. The one detail I told about opnsense bui= lding is that I use pkg 2.3.1_1, as is the version that ships on the imag=
es and the port is on opnsense ports github repo. Other than that, regula=
r FreeBSD 15.1-R. I download the github zip file for ports, run make inst=
all on the pkg port and lock the package after. Used to work fine, can th=
is be the culprit? I got some messages after upgrade about libpkg not the=
same version or database version. But I then installed it all over and n=
o more of that.
=20=20
So I tried to clone the repo today on a Linux Desktop here (Linux = Mint), all fine. Same network, same servers, same ISP. I then tried on an=
old amd64 box running FreeBSD 15.1-R. Again, fine. Both my arm64 failed,=
even after fresh install. Can the pkg port be the one? Before I forget, =
both the arm64 run on magnetic HDDs over USB adapter and all other run on=
SSDs over SATA. Would this make this all happen?
=20=20
After several tries (3 days) I got to clone the repo and got to st=
art building. Ports used to take 24h and now runs for 5 days. So I got in=
fo about llvm changes and got the config to try to solve this (/etc/src.c= onf), and my point is just that, I need to build vanilla FreeBSD 15.1-R o=
n arm64 (just using older pkg, I can try to not change now if people thin=
k may help) so I can have world and kernel using llvm options from 14.X.
=20
=20Start distinct subject from the main point of your post:
=20
=20Were you successful at significantly decreasing that "5 days" time
frame? If yes, you may want to be very explicit about what you changed
that made the difference.
=20
=20The official arm64 build server build times suggests that 14.4 now ta=
kes
longer than 14.3 did --and that 15.1 takes even longer, as I remember.
=20
=20("Suggests" is tied to: none of the examples are re-runs of the exact
same ports tree content --or, mostly, the same package subset. Comparin=
g
the fairly uncommon attempts to rebuild everything tend to be what I
look at for such comparison/contrast activities.)
=20
=20End distinct subject.
=20
Hello=20Mark and others, I will try to bring some answers here.
At first, I thought it was faster. But now its looking like a solid "no".
The ports build phase is running since Aug 03 22:44 GMT -3, and it looks =
to be 75% done. So quite close. On the other hand, git clone looked bette=
r (I really can't tell why, but the timed out rate when using clone and d=
epth 1 was really low). The world and kernel build time for Orange Pi 5 P=
lus 16GB was 3.5 hours. Again, I use HDD over USB not SSD.
My future plans are try to crosscompile on AMD64, that was problematic be=
fore and I will try to find out if its viable now. Build times were far b=
elow 2 days and now I have 4 days and counting.
Thanks as usual for the help. When its over I can post here the actual ti=
me if there is people interested on the info.
matheus
--aafbdc66-34b5-4160-9698-0d88a3170554-1
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE html><html><head><meta http-equiv=3D"Content-Type" content=3D"t= ext/html; charset=3Dutf-8"></head><body><div><br>August 1, 2026 at 2:25 P=
M, "Mark Millard" <<a href=3D"mailto:
marklmi@yahoo.com?to=3D%22Mark%20= Millard%22%20%3Cmarklmi%40yahoo.com%3E" target=3D"_blank" tabindex=3D"-1"= >
marklmi@yahoo.com</a>> wrote:</div><blockquote>On 8/1/26 08:46, Nenhu= m_de_Nos wrote:<br><blockquote>August 1, 2026 at 10:47 AM, "Karl Denninge=
r" <
karl@denninger.net<br> <mailto:
karl@denninger.net?<br> to=3D%22= Karl%20Denninger%22%20%3Ckarl%40denninger.net%3E>> wrote:<br> <br> =
The problem appears to be at this point:<br> <br> Cloning into '/= usr/src'...<br> fatal: unable to access '<a href=3D"
https://git.freeb= sd.org/src.git/" target=3D"_blank" tabindex=3D"-1">
https://git.freebsd.or= g/src.git/</a> <<a href=3D"
https://git.freebsd.org/src.git/%3E':" targ= et=3D"_blank" tabindex=3D"-1">
https://git.freebsd.org/src.git/>':</a> = Resolving timed out after 300226 milliseconds<br> <br> I do not know =
why, however, because my "base" source git clone has<br> in its confi=
g the following:<br> <br> [remote "freebsd"]<br> =C2=A0 =C2=A0 =
=C2=A0 =C2=A0 url =3D <a href=3D"
https://git.freebsd.org/src.git" target= =3D"_blank" tabindex=3D"-1">
https://git.freebsd.org/src.git</a> <https= ://<br> git.freebsd.org/src.git><br> =C2=A0 =C2=A0 =C2=A0 =C2=
=A0 fetch =3D +refs/heads/*:refs/remotes/freebsd/*<br> <br> And it wo= rks.<br> <br> I have several worktrees checked out against that since=
I build for<br> 14-STABLE and 15-STABLE at this point for distributi=
on to machines<br> running ARM64 and AMD64.<br> <br> My latest "g=
it pull --ff-only" to so-update was this morning.<br> <br> On 8/1/202=
6 09:36, Nenhum_de_Nos wrote:<br> <br> Hi,<br> <br> I hav=
e been fighting this issue for 2 weeks now, I guess its time to ask here.=
The problem:<br> <br> root@arm_build_hd:/usr # rm -rf src/<br> =
root@arm_build_hd:/usr # /root/git/freebsd_config/freebsd_git/src_= git_checkout_15.1R.sh<br> Cloning into '/usr/src'...<br> = fatal: unable to access '<a href=3D"
https://git.freebsd.org/src.git/" tar= get=3D"_blank" tabindex=3D"-1">
https://git.freebsd.org/src.git/</a> <<=
a href=3D"
https://git.freebsd.org/src.git/%3E':" target=3D"_blank" tabind= ex=3D"-1">
https://git.freebsd.org/src.git/>':</a> Resolving timed out = after 300226 milliseconds<br> root@arm_build_hd:/usr # /root/git/= freebsd_config/freebsd_git/src_git_checkout_15.1R.sh<br> Cloning =
into '/usr/src'...<br> remote: Enumerating objects: 4850942, done= .<br> remote: Counting objects: 100% (384686/384686), done.<br> =
remote: Compressing objects: 100% (29598/29598), done.<br> =
error: RPC failed; curl 56 OpenSSL SSL_read: OpenSSL/3.5.6: error:0A0001= 26:SSL routines::unexpected eof while reading, errno 0<br> error:=
73 bytes of body are still expected<br> fetch-pack: unexpected d= isconnect while reading sideband packet<br> fatal: early EOF<br> =
fatal: fetch-pack: invalid index-pack output<br> root@arm= _build_hd:/usr # /root/git/freebsd_config/freebsd_git/src_git_checkout_15= .1R.sh<br> you have mail<br> Cloning into '/usr/src'...<b=
fatal: unable to access '<a href=3D"https://git.freebsd.org/sr=
c.git/" target=3D"_blank" tabindex=3D"-1">
https://git.freebsd.org/src.git= /</a> <<a href=3D"
https://git.freebsd.org/src.git/%3E':" target=3D"_bl=
ank" tabindex=3D"-1">
https://git.freebsd.org/src.git/>':</a> Resolving=
timed out after 300040 milliseconds<br> root@arm_build_hd:/usr #=
cat /root/git/freebsd_config/freebsd_git/src_git_checkout_15.1R.sh<br> =
git clone --branch releng/15.1 <a href=3D"
https://git.freebsd.org/= src.git" target=3D"_blank" tabindex=3D"-1">
https://git.freebsd.org/src.gi= t</a> <<a href=3D"
https://git.freebsd.org/src.git%3E" target=3D"_blank=
" tabindex=3D"-1">
https://git.freebsd.org/src.git></a> /usr/src <br> =
root@arm_build_hd:/usr # <br> <br> On two Orange Pi 5 Plus=
16GB I have FreeBSD 15.1-R installed so I can build opnsense for arm64 (=
this is not relevant for the problem, except for one detail I will tell l= ater). This box above had 14.3-R running fine, all build times were fine,=
git worked perfectly. Since I needed to upgrade to 15.1-R, git stopped w= orking as it was expected (and presented above).<br> <br> I looke=
d for DNS reasons, changed DNS server, local server, remote server, kept =
ping github.com while cloning, no good. I started using the upgraded box,=
then I got another similar arm64 box and installed fresh 15.1-R, everyti=
me the same issue. The one detail I told about opnsense building is that =
I use pkg 2.3.1_1, as is the version that ships on the images and the por=
t is on opnsense ports github repo. Other than that, regular FreeBSD 15.1=
-R. I download the github zip file for ports, run make install on the pkg=
port and lock the package after. Used to work fine, can this be the culp=
rit? I got some messages after upgrade about libpkg not the same version =
or database version. But I then installed it all over and no more of that= .<br> <br> So I tried to clone the repo today on a Linux Desktop =
here (Linux Mint), all fine. Same network, same servers, same ISP. I then=
tried on an old amd64 box running FreeBSD 15.1-R. Again, fine. Both my a=
rm64 failed, even after fresh install. Can the pkg port be the one? Befor=
e I forget, both the arm64 run on magnetic HDDs over USB adapter and all = other run on SSDs over SATA. Would this make this all happen?<br> <br> =
After several tries (3 days) I got to clone the repo and got to sta=
rt building. Ports used to take 24h and now runs for 5 days. So I got inf=
o about llvm changes and got the config to try to solve this (/etc/src.co=
nf), and my point is just that, I need to build vanilla FreeBSD 15.1-R on=
arm64 (just using older pkg, I can try to not change now if people think=
may help) so I can have world and kernel using llvm options from 14.X.</= blockquote><br>Start distinct subject from the main point of your post:<b= r><br>Were you successful at significantly decreasing that "5 days" time<= br>frame? If yes, you may want to be very explicit about what you changed= <br>that made the difference.<br><br>The official arm64 build server buil=
d times suggests that 14.4 now takes<br>longer than 14.3 did --and that 1=
5.1 takes even longer, as I remember.<br><br>("Suggests" is tied to: none=
of the examples are re-runs of the exact<br>same ports tree content --or=
, mostly, the same package subset. Comparing<br>the fairly uncommon attem=
pts to rebuild everything tend to be what I<br>look at for such compariso= n/contrast activities.)<br><br>End distinct subject.<br><br><br></blockqu= ote><div><br></div><div>Hello Mark and others, I will try to bring some a= nswers here.</div><div><br></div><div>At first, I thought it was faster. =
But now its looking like a solid "no".</div><div><br></div><div>The ports=
build phase is running since Aug 03 22:44 GMT -3, and it looks to be 75%=
done. So quite close. On the other hand, git clone looked better (I real=
ly can't tell why, but the timed out rate when using clone and depth 1 wa=
s really low). The world and kernel build time for Orange Pi 5 Plus 16GB =
was 3.5 hours. Again, I use HDD over USB not SSD.</div><div><br></div><di=
My future plans are try to crosscompile on AMD64, that was problematic = before and I will try to find out if its viable now. Build times were far=
below 2 days and now I have 4 days and counting.</div><div><br></div><di= v>Thanks as usual for the help. When its over I can post here the actual =
time if there is people interested on the info.</div><div><br></div><div>= matheus</div></body></html>
--aafbdc66-34b5-4160-9698-0d88a3170554-1--
--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to
news-admin@muc.de
--- Synchronet 3.22a-Linux NewsLink 1.2