## Problems I ran into, and the solutions that worked for me. ## Building wayland on Scientific Linux 6.x -- 05/24/14 ---------------------------------------------------- Downloaded wayland 1.5.0 and ran configure, got these errors: 1) ERROR: checking for FFI... no configure: error: Package requirements (libffi) were not met: No package 'libffi' found At first this didn't make sense, because 'yum install libffi' said it was already installed. But what was installed was the runtime libs, not the development files. So... SOLUTION: yum install libffi-devel 2) ERROR: configure: error: Can't find expat.h. Please install expat. Similar problem as (1); runtime 'expat' libs were installed, but not the devel stuff. (e.g. include files) SOLUTION: yum install expat-devel With that, wayland built and installed OK with './configure && make && make install' Installing Ubuntu 12.04 server on SunFireVZ AMD64 systems -- 05/24/14 --------------------------------------------------------------------- Ran into these gotchyas during the install: 1) During final stages of the install, grub failed. Looking in the Ctrl-Alt-F4 error console, I saw the errors about: /usr/sbin/grub-setup: error: hd0 appears to contain a udf filesystem which isn't known to reserve space for DOS-style boot. Installing GRUB there could result in FILESYSTEM DESTRUCTION if valuable data is overwritten by grub-setup (--skip-fs-probe disables this check, use at your own risk). The command the installer was running is shown as: chroot /target grub-install --no-floppy --force /dev/sda So I tried adding a "--skip-fs-probe" to that command, but 'grub-install' does not support that flag, only grub-setup does. So couldn't force grub to overwrite it. Turns out the problem was just some leftover crap on sector zero of the drive that grub was all bitchy about. There had previously been a Windows XP install on the machine, and grub apparently was noticing that data, and didn't want to overwrite it. I didn't need that stuff and wanted to overwrite it, so a simple solution was to just fill that first sector with zeros. I opened a shell in one of the spare linux consoles (Ctrl-Alt-F2), then ran: SOLUTION: dd if=/dev/zero of=/dev/sda bs=512 count=1 This overwrites sector zero with 512 zeroes at the head of /dev/sda so that the grub installer would see zeroes instead of the old data. This worked, because I was then able to go back to the installer screen (Alt-F1) and from the final menu, was able to retry the grub install, and it completed the install just fine and rebooted the new OS. 2) Made a dd copy of the successful install harddisk on another harddrive of the same size, and it booted OK when I changed the hostname/IP info, but networking wouldn't come up. (During boot there was a long wait while it tried to wait for networking to come up.. after 60 seconds it gave up, and 'ifconfig' showed no network interface up) Turns out on the first install, the installer found the ethernet device as eth0, but on this machine, it saw it as eth2. I found this out when I tried to manually bring up the interface with: ifconfig eth0 up 192.168.2.38 ..and it gave errors about no such device. SOLUTION: Use 'ifconfig -a' to see what interfaces are available, and found eth2 and eth3. Then modified the /etc/network/interfaces file, changing the "eth0" reference to "eth2", and was able to test by restarting networking: /etc/init.d/networking stop /etc/init.d/networking start It seemed weird this was happening, because the two machines are exactly alike; each has two ethernet interfaces built into the mother board. Perhaps some leftover settings from the clone made it think there were two /new/ ethernet ports, so it numbered them eth2 and eth3.