We bought a few of the newish Mac Mini “servers” at work – with two hard drives. They came pre-installed with Mountain Lion and the disks were configured independently – no RAID. There’s also no optical drive (fine by me). Putting a Mountain Lion installation (InstallESD.dmg) on a USB stick produced an unbootable volume. Apparently, the retail copy of the Mountain Lion installer doesn’t contain necessary drivers for the newish Minis, so it didn’t boot. Booting into the “recovery partition” on the Mini didn’t permit creating a RAID, since the recovery partition is on the mounted volume.
My solution was to copy the recovery partition to the USB volume, boot off that, configure the RAID, and install from that.
To do that, boot to the installed OS. Enter the following to enable Disk Utility’s debug menu – this will allow you to see/work with the recovery partition:
defaults write com.apple.DiskUtility DUDebugMenuEnabled -bool true
Open Disk Utility. Make sure the USB volume has a Mac (HFS+) filesystem on it. Mount the recovery partition (yes). Click the restore tab and make the source the recovery partition and the destination the partition on the USB volume. Trying to drag the USB device itself to the restore field doesn’t work. Trying to “restore” if the recovery partition isn’t mounted first doesn’t work, either.
Hope this is useful to someone Googling.
This only applies to running ISC-DHCPD in a chroot environment on FreeBSD. (The port option is “PARANOIA” and is default).
Ensure /usr/local/etc/dhcpd.conf has the following:
Add this to /etc/rc.conf
Add this to /etc/syslogd.conf
Create the log file:
touch /var/log/dhcpd.log && chmod 640 /var/log/dhcpd.log
And restart services:
service syslogd restart
service isc-dhcpd restart
You’ll also want to rotate the logs. Here’s an example.
# This will rotate daily at 5 AM, keep 7 copies, and bzip2 the older logs.
/var/log/dhcpd.log 640 7 * $D05 JC
See the man pages for syslogd and newsyslog.conf for more information.
I can’t believe I wasn’t doing this sooner. I wrote a simple, quick shell script to confirm you’re serious about restarting/shutting down a server and log the reason for it with the date/time and user that initiated shutdown.
Get it from my GitHub
As big of a geek as I am, I still don’t have a smart phone. In fact, my cell phone is a crappy flip-phone that my employer provides and I use Google Voice quite heavily for things like SMS.
I entertain the idea of getting something every now and then, but this is why I never do (and still refuse to):
So I’m looking at a minimum monthly charge of ~$100. My cable (extended w/ dvr and HD) + broadband (modest 35 Mbps) is cheaper than that even after taxes!
Update 2012-09-08: I suggest Insync, which offers a tray icon and Google Docs support.
Unfortunately, Google still hasn’t delivered a native Google Drive client for Linux. Until then, we have Grive by Nestal Wan.
Grive by itself doesn’t have any sort of GUI or notifications. I’ve hacked up a script to be called via cron that will run Grive to sync and display desktop notifications of what happened.
Here’s a couple of screenshots:
I’ve always enjoyed QNX’ Photon interface, especially the window manager decorations. For every window manager I’ve ever ran, I’ve created themes based on that look.
There’s a rather dated theme on GNOME-Look that brings it to Metacity, but with right-side buttons.
I’ve modified this theme for left controls:
I have a MacBook Pro from my employer that I triple boot Mac OS X, Ubuntu, and Windows 7 on (see rEFIt)
Last year, I wasn’t easily able to upgrade to Lion because the installer refused to install – apparently because I had Linux partitions. It failed with a vague error “This disk cannot be used to start up your computer.” Without much investigation, I decided to just rebuild the system.
This year, I wanted to upgrade from Lion to Mountain Lion but encountered the same issue. Ugh. Pulling off all three partitions, installing Mountain Lion, then re-creating them and restoring data didn’t sound too appealing. Apple mentions this in this article, but that method doesn’t work, unfortunately.
Solution: I booted to a Linux live disk and used GParted to shrink the size of the Mac (HFS+) partition slightly. As Apple’s document mentions, shrinking it by at least 128MB should work. After that, I was able to install.
Hope this helps someone Googling this.
I migrated the various services I hosted at Linode to my home FreeBSD server over the weekend. I love Linode, but wanted to utilize my humble home server more.
I decided to migrate my various domains’ e-mail to Google Apps. This was a hard transition for me. I’ve hosted my own e-mail for 10 years or so now, even from my home cable connections. GMail is excellent. I’ve started using it exclusively by using its POP3 checking functionality to aggregate all my various accounts into a single location and use GMail’s labels and filters to clean things up. Google Apps is free for non-profits for up to 10 accounts (and you can set a catch-all address). No regrets so far and really enjoying it.
So one thing I’ve migrated is DNS to my own server, where it was previously Linode. This would be painless, but my registrar, Netfirms, sucks. One thing after another there.
I registered two domains with them two years ago and transferred my “hewbert.com” domain to them at the same time. I logged into the control panel and “hewbert.com” showed that it wasn’t transferred, despite whois confirming they were the registrar. My other two domains (this one included) showed I was overdue for renewal by 277 days, despite whois confirming the accurate truth – I have 77 days left! So I did a chat support, where I was met with a scripted, incompetent sounding person that ultimately had to “file an escalation.” The two domains were sorted out.
In the meantime, I started changing my DNS records at the registrar. I previously had ~6 nameservers set for each domain. I started with this domain – changing it to a new set. As soon as I changed them, their control panel chopped off all but two of them. I was left with only two nameservers with no way to change. Chat support again. They say “we limit you to two nameservers. You need to contact us if you want more.” Umm… okay, I’m contacting you. They direct me to a completely separate site that looks like it was made 15 years ago, but I was able to modify the nameservers as I wanted there. I asked the agent twice if I’d be able to update all of my domains there and she said yes. After logging in, I could only modify one domain. Ugh.
I’m going to add to this as I find them. These are server-specific, but likely apply to certain desktop situations as well.
Upgrade fails because /var/lib/defoma/fontconfig.d/ isn’t empty
This pisses me off. Remove /var/lib/defoma/fontconfig.d before upgrading.
If you already started an upgrade and it bails, complaining about fontconfig.d, remove that directory and restart the upgrade. If the upgrade says no releases are available, do apt-get install -f
Link to bug report
libvirt lost my VMs
virsh list shows up empty.
Solution: Remove the <console> entries from the XML files in /etc/libvirt/qemu/*.xml
I haven’t looked into this yet, but I’ll edit this post once I do.
Link to bug report
I wrote up a quick guide on how to install and use Policyd v2 (cluebringer) to limit mail with Postfix. This guide is geared towards Ubuntu, but could easily be implemented elsewhere.
Check it out at /howto/policyd/