Hi Samuel, > I don't know what this cfarm is about,
CFarm is the GCC Compile Farm. It is where FOSS developers go to test their software on different pieces of hardware and software. Hurd is a popular choice on the CFarm because it has a lot of personality. Also see <https://gcc.gnu.org/wiki/CompileFarm>. Jeff On Tue, Mar 10, 2026 at 8:20 AM Samuel Thibault <[email protected]> wrote: > Hello, > > I don't know what this cfarm is about, but: > > - FI, mosh should build fine > - I'm surprised by the 6G limitation on 64b, I'm running a 16G VM > without a problem. > - tmpfs shouldn't be used, see > https://darnassus.sceen.net/~hurd-web/open_issues/tmpfs/ > - the 128GB limitation on 32b can be avoided by using rumpdisk, see > https://darnassus.sceen.net/~hurd-web/faq/2_gib_partition_limit/ > https://darnassus.sceen.net/~hurd-web/hurd/rump/rumpdisk/ > > Samuel > > Jeffrey Walton, le mar. 10 mars 2026 08:12:53 -0400, a ecrit: > > FYI... > > > > Original message at <[1] > https://lists.tetaneutral.net/pipermail/cfarm-users/ > > 2026-March/001542.html>. > > > > ---------- Forwarded message --------- > > From: Luke Yasuda via cfarm-users <[2][email protected]> > > Date: Tue, Mar 10, 2026 at 7:11 AM > > Subject: [cfarm-users] cfarm{431,432}: hurd-{amd64,i386} > > To: Cfarm Users <[3][email protected]> > > Cc: Luke Yasuda <[email protected]> > > > > Hi all, > > > > The two Debian GNU/Hurd VMs, cfarm431 (hurd-amd64) and cfarm432 > > (hurd-i386) are available now. But they need some "special lovin'" as > > the highest uptime record on munin is 5.61 days so far. On the spirit of > > sharing notes and experience and hoping someone on the mailing list > > might have some insights: > > > > common issues on hurd-*: > > 1. ntp or chrony or any other NTP implementation (that I know of) won't > > work. If you notice that the time is skewed, email me so I can correct > > it manually. > > 2. we have very limited CPU and RAM resources (also see below). Large > > builds can take days. Coordination among users may be necessary. We > > should use this mailing list to coordinate when the time comes. If > > posting on this list is "too loud" etc (if anyone complains), I can > > create a mailing list under @lists.jing.rocks (mailman3 managed) > > 3. upgrading hurd and glibc might crash and even cause the system to be > > unbootable. So, from now on, I'll make a snapshot before upgrading them. > > Basically I check for package upgrades about once a week. > > 4. I absolutely cannot get IPv6 to work. I wrote a custom init script > > (/etc/init.d/networking-real) because ifupdown doesn't implement IPv6 on > > hurd yet. > > 5. mosh is not available. fstrim is not available. As well as so many > > other packages. > > 6. hurd does not support having 2 or more IPv4 addresses or configuring > > two or more netif (not including loopback "lo"). > > 7. sshd may not be working at all when it's started on boot, needs to be > > restarted on tty by hand: > > > > Mar 10 10:37:29 cfarm432 sshd[860]: error: ppoll: Computer bought the > > farm > > Mar 10 10:38:00 cfarm432 last message repeated 350578 times > > Mar 10 10:38:02 cfarm432 last message repeated 25942 times > > > > > > hurd-amd64: > > 1. only 1 core is available. I cannot get the SMP kernel > > (gnumach-image-1-amd64-smp) to boot, it just resets itself after grub > > loads gnumach. I don't intend to debug it due to the lack of skills > > 2. only 6GB RAM is available. Again, it resets itself at boot time when > > there is more than about 8GB system memory > > 3. /home is 500GB large (not an issue per se). After a system crash, it > > may take 15 min or so for e2fsck to check and fix the file system. > > 4. recently, every 3~5 days or so, for some reason, /hurd/tmpfs will > > crash and also bring down the networking. > > > > hurd-i386: > > 1. similar to hurd-amd64, only 1 core is available. I cannot get the SMP > > kernel (gnumach-image-1-486-smp) to boot > > 2. only 3GB RAM is available. I cannot get the PAE kernel > > (gnumach-image-1-486-pae) to boot > > 3. /home is only 128GB large. hurd-i386 cannot recognize a disk larger > > than 128GB, any disk larger than 128GB will just appear to be 128GB, > > which seems to be an ancient C/H/S addressing limitation existed in > > windows 2000 if I remember it correctly. So, the space is tight and > > remember to clean up after yourself, dear friends. > > > > Cheers, > > -- > > Luke Yasuda > > GPG Fingerprint: 4E09 8D19 00AA 3F72 1899 2614 09B3 316E 13A1 1EFC > > _______________________________________________ > > cfarm-users mailing list > > [4][email protected] > > [5]https://lists.tetaneutral.net/listinfo/cfarm-users > > > > References: > > > > [1] > https://lists.tetaneutral.net/pipermail/cfarm-users/2026-March/001542.html > > [2] mailto:[email protected] > > [3] mailto:[email protected] > > [4] mailto:[email protected] > > [5] https://lists.tetaneutral.net/listinfo/cfarm-users > > > -----BEGIN PGP SIGNATURE----- > > > > iHUEARYIAB0WIQQUNK5y7dM5LGmlOjiPRdGe/wwPKwUCaa/8DAAKCRCPRdGe/wwP > > K/QGAP9/xZcqHW6V5R77U92WGRB96Pua8k33JFBa+Z5L1qdfVwD7Bhs/6MDSfwqQ > > yvWhm52pBe5ZwSsKeyy1kjtgj4Djcg0= > > =YJgL > > -----END PGP SIGNATURE----- >
