Gentoo graybeards (you know who you are) are undoubtedly familiar with the mirrorselect utility. This simple program (script, actually) attempts to determine which Gentoo mirror is the fastest for you. By specifying a -D option, it will even try to download a 100 KB file to measure the throughput.
After tearing out my hair for the past few days, finally I found the reason why my Gentoo PV VM can’t boot:
_ I had changed CONFIG_VMSPLIT from its default value of 3G/1G to 2G/2G _
I rather naïvely thought that, because the PV VM is meant to be a vFirewall, I should allocate more memory space to the kernel. (After all, netfilter/xtables is a kernel function, not a userspace function.)
As a result, attempting to start the Gentoo PV VM just… failed. No error messages, no logs, nothing. XenServer acted as it’s starting up the VM (the progress bar increases), then the process is complete. But the VM is still in the ‘off’ stage.
Today, I booted from a ‘last-known-bootable’ VM kernel, and worked my way down from there. Page-by-page I changed the kernel config, do a make && make modules_install, and so forth. To cut a long story short, I finally found that any setting for CONFIG_VMSPLIT besides 3G/1G will result in a non-bootable VM.
Oh well. It’s not essential, anyways. So I returned the value to its default of 3G/1G, and now I’m a happy dawg :)
First, some explanation re: ‘stage3.5’. I had planned on calling this a ‘stage4’, but there are some confusion regarding the term ‘stage4’, even amongst Gentoo users. To prevent howls of protests that might possibly be raised by Gentoo graybeards, I decided to call this a ‘stage3.5’ instead.
Okay, what the heck is a ‘stage3.5’? The numbering implied that it’s a continuation of ‘stage3’. And in fact, it is. It’s a tarball containing:
A customized /etc/make.conf
A customized /etc/portage/ (accepted keywords, package-specific use flags, rsync excludes, etc)
An updated portage
A @world that’s gone through emerge -uDN and revdep-rebuild
In other words, all the stuffs just before emerging the kernel sources.
I have just read in Washington Post that The White House just lost email service due to server outage. Now that is ridiculous, indeed! Can’t White House set aside 50’000 USD to pay for a qualified IT OPTIMIZER? With an IT Optimizes in-house, such snafu won’t happen for he/she will ensure that all infrastructure - including email servers - run optimally. If you don’t know what an IT OPTIMIZER is, check out my manifesto here: http://ping.fm/cqhHN
I truly believe that the so-called ‘IT Age’ has arrived. Or should I say, has crept upon us. Even telecom companies are now using a single, converged IP network to deploy their telecom backbone. For example: XL. They no longer use separate platforms for telephony and data, but they use the same IP network for all their needs. Amazing.
The Cisco Unified Personal Communicator *is* interesting. Think of it as a combination of Instant Messenger, Skype (with video), and PBX front-end. Not just that, you can associate your “contact point” with your desktop, your office phone, and your cellphone. Thus, if someone calls your contact point, you can receive your call anywhere, even on the road!
Cisco UC is apparently suitable for really large enterprises only. To be effective, the Cisco UC should be able to replace high-cost ops like cross-country travel, inter-city inter-site phone call, and so on. Further, you need very good bandwidth. My office, though handling a lot of money, is much more similar to SMBs in its networking needs.
Sometimes sleep is very hard to get. Like tonight. I have been lying here for one hour, my eyes *are* feeling heavy, yet sleep seems to be avoiding me. So here I am, lying in the dark, thinking about the universe’s great secrets. Not that I am a philosopher or something, mind you, but because the universe is so big I expect my brain to be overcome by it and fall into a deep knock out. Unfortunately, lately that doesnt help either. For I know the answer is 42.