Jacob Bachmeyer via cfarm-users writes:
>Which is why I am looking for solutions that also give us conventions for
>"politely" running other jobs like CI or fuzzing,
Fuzzing is essentially identical to mining in terms of what it does to a host,
even if it's for good (non-klepto) purposes. I thi
Jing Luo via cfarm-users writes:
>The ampere altra (aarch64) system that cfarm424~426 live on can accept
>another 1 or 2 "small" vm. I'm thinking either: two vm, each with 16 cores +
>16GB RAM + 1TB /home, or just one vm with 32 cores + 32GB RAM + 2TB /home.
Just for reference, I'm currently run
CFarm Annoucements via cfarm-announces
writes:
>We are happy to announce the immediate availability of cfarm423, cfarm424,
>cfarm425, and cfarm426, all of which are aarch64 systems.
Just a thought here, we're kinda inundated with Debian (both cfarm and non-
cfarm) but there's only a single BSD
Jacob Bachmeyer via cfarm-users writes:
>512-bit RSA is definitely breakable and should not be used for a long-term
>key. 768 bits is also too short; 1024 is currently marginal and definitely
>not suitable for a high-value target, but impersonating a cfarm host will not
>get an attacker much oth
Thorsten Glaser via cfarm-users writes:
>It’s not, for example you can use double quotes "…" only either around or
>inside accent-gravis-style command substitutions `…` but not both at the same
>time.
Oh, I meant in my case where everything was `` to begin with and got seek-and-
replaced with $(
Denis Ovsienko via cfarm-users writes:
>I ran into this particular thing after following the advice of shellcheck and
>replacing `` with $(), which promptly broke the script on Solaris 10, which
>is supposed to be POSIX-compliant, at least in the shell department. As far
>as I remember, eventual
Thorsten Glaser via cfarm-users writes:
>So, please keep it that way ;-)
+1. Having access to a very diverse range of compilers and build
environments, even if some are broken-by-design from the vendor, is extremely
useful both to shake out bugs and for regression testing on older systems.
x86-
Bruno Haible writes:
>Look at the PATH column in
>https://gitlab.com/ghwiki/gnow-how/-/wikis/Platforms/Configuration.
Ah, that fixed it, thanks, I was missing /usr/ccs/bin.
Speaking of /usr/ccs/bin, it's nice to have an excuse to haul out my Version 7
Unix manuals again.
(OK, it's not quite th
Jonathan Wakely writes:
>>cfarm210 is running an ancient version of SSH that requires all sorts of
>>algorithm downgrades to work. This isn't a big deal security-wise, but it
>>does mean adding a pile of overrides to the client-side SSH.
>
>That's an out-of-support version of Solaris, there's a
Bruno Haible writes:
>cfarm119 has two compilers installed:
> - A GCC that works, in /opt/freeware/bin.
> - An IBM derivate of clang, named ibm-clang and ibm-clang++,
>in /opt/IBM/openxlC/17.1.1/bin. But it is broken (miscompiles various
>things here and there).
Ah, OK. According to m
Playing with my own build-on-everything script I've found some issues with
several systems, when doing things manually I've just skipped over them but
the script is less forgiving...
cfarm23 has an ancient clang install that dies with an internal error trying
to generate code.
cfarm112 has a brok
Jing Luo via cfarm-users writes:
>Suggestion for the music to play when I work on the servers is very welcome.
Sunn O))) - Live at The Mayan. Then you could see if you can finish the task
before they get around to their second note.
Peter.
___
cfarm-
Paul H. Hargrove writes:
>Have a look at `ControlPath`, `ControlPersist` and similar in the ssh man
>page.
Very useful, thanks! Looks like others have run into, and already solved,
this problem in the past.
Peter.
___
cfarm-users mailing list
cfarm-u
Peter Gutmann via cfarm-users writes:
>I was hoping to open one SSH connection, run a sequence of commands, then
>disconnect, in order to avoid hammering the servers with repeated SSH
>handshakes.
Forestalling the inevitable responses :-), I'm aware of the use of '... &&
Martin Guy via cfarm-users writes:
>Funny you should ask. I just implemented a set of scripts to perform
>operations on all compile farm machines including installing build
>dependencies under $HOME, configuring with different flags on some
>architectures, compiling a prog and running its regress
Simon Josefsson writes:
>I thought this usage was against the recommended use of the cfarm, but I
>cannot find a reference now -- was the acceptable use policy modified?
It's just scripting what's otherwise done manually, it's taking the same
commands that I'd normally type out and putting them i
Is anyone doing CI-style automated deployment across multiple cfarm hosts, or
more generally on systems where all you've got to play with are SSH, sh/bash,
and a compiler? Conventional approaches seem to require installing CI agents
on each target system (and Docker, and cloud access, and ...), I
Martin Jambor via cfarm-users writes:
>Which might be the right general rule to do in general. Something like: "If
>your home does not contain a file younger than 6 (or 9?) months, its contents
>can be purged in its entirety by admins if disk ever runs low on a machine.
>The bigger the home, the
Gregor Riepl via cfarm-users writes:
>It's because it relies on SHA-1, and SHA-1 is known to be broken.
It's breakable with a considerable amount of effort for static data. Attacking
it in SSH auth would require the ability to break it in real-time or close to
it, which no-one has come close to
Jacob Bachmeyer via cfarm-users writes:
>This is a pet peeve of mine: unless you have a citation for an actual viable
>attack on RSA as used in SSH, or perhaps on the protocol SSH uses for RSA-
>based authentication, this is *not* insecure at all and those changed
>defaults indicate that either
Jonathan Wakely writes:
>If you have this problem with a tarball
It's not tar, see my other post, it's assorted broken archive formats some
dating back to MSDOS. I suggested the 'make touch' solution because it pretty
much works no matter where the breakage lies.
Peter.
__
Bruno Haible writes:
>If time zones do matter, either your packing program or your unpacking
>program is buggy.
It's not buggy, it's just a poorly-designed archive format. That's why I used
the 'make touch' workaround, which conveniently also gets around wrong-system-
time problems, time-zone-s
Paul Eggert writes:
>On 8/20/23 21:19, Peter Gutmann via cfarm-users wrote:
>> As someone in a time zone 12-24 hours ahead of most of the systems I use
>
>? Time zone shouldn't matter.
If you bundle something up on a system that's (say) 12 hours ahead of the
target s
Bruno Haible via cfarm-users writes:
>Because in the current state, it means that after creating a tarball of any
>package that uses a 'configure' script, I need to wait 42 minutes until I can
>build it.
As someone in a time zone 12-24 hours ahead of most of the systems I use I run
into this pro
I was running an alt build config on gcc112 and noticed that the clang install
there is b0rked:
> cat >> hello.c
#include
int main( void )// A lie...
{
puts( "Hello world" );
return( EXIT_SUCCESS );
}
^D
> clang hello.c
In file included from hello.c:1:
In file included from /usr/include
Pierre Muller via cfarm-users writes:
>LLVM ERROR: out of memory
Building LLVM from source consumes ridiculous amounts of memory, particularly
if you build across multiple CPUs. If you're doing a parallel build you may
want to limit it to 1/2 to 2/3 of the total number of cores/CPUs/whatever.
Paul Eggert via cfarm-users writes:
>No, on traditional Solaris 10, traditional 'make' is /usr/ccs/bin/make (part
>of the SUNWsprot package) or /usr/xpg4/bin/make (part of the SUNWxcu4t
>package). They are identical executables. Although neither is required as
>part of operating system core, both
Paul Hua via cfarm-users writes:
>Would you please provide test.c for debugging.
Sure:
cat > test.c
#include
^D
Peter.
___
cfarm-users mailing list
cfarm-users@lists.tetaneutral.net
https://lists.tetaneutral.net/listinfo/cfarm-users
Bruno Haible via cfarm-users writes:
>On both machines, the clock is 8 hours behind reality.
Being in a country that's ahead of everyone else on earth's time zone, I've
become used to "clock skew detected" almost as an extra motd. A quick
workaround if you're using zipped archives is to add -D
cfarm-announces writes:
>Feel free to report any issues
First issue :-). Looks like clang is present but not set up:
> gcc test.c
> clang test.c
In file included from test.c:1:
In file included from /usr/include/stdlib.h:394:
In file included from /usr/include/sys/types.h:227:
In file include
30 matches
Mail list logo
