Disqus - Latest Comments for phrenologist

Re: Libre Computer ROC-RK3328-CC Raspberry Pi Alternative

phrenologist — Mon, 25 Jan 2021 01:09:52 -0000

This board has terrible linux support. I just spent an entire weekend trying to get anything besides the Libre-supplied Ubuntu 18.04 or Debian 9 to work. Current versions of Armbian don't work, and this is not a supported board so it's not clear if they ever will.

There are other similar boards out there with much better linux support, so I suggest you try one of those instead of wasting time with this one.

Re: HollenbackDotNet - Roku M500 Automation

phrenologist — Thu, 14 Nov 2019 12:39:33 -0000

Thank you. You have no idea how gratifying comments like yours are. Most of the time my posts on this wiki are just telling in to the void.

Re: HollenbackDotNet - Lets Encrypt Setup

phrenologist — Tue, 01 Aug 2017 15:42:07 -0000

Yes, I just set up a cronjob to check the file every 5 minutes and fix the perms if they changed. From /etc/crontab:

*/5 * * * * root chmod 640 /etc/letsencrypt/archive/<hostname>/privkey*.pem

Re: Review of Airbnb's SRE Interview Process

phrenologist — Fri, 17 Aug 2012 13:46:12 -0000

You guys are making baby kittens cry right now.

Re: Review of Airbnb's SRE Interview Process

phrenologist — Wed, 15 Aug 2012 21:19:47 -0000

Ha. Based on our experiences using zookeeper as a backend for a massively parallel ssh execution engine I would say zookeeper may not be the best solution for everything. We've had massive scalability problems with it, but to be truthful I don't know if that's just because what we did with it turned out to not be a good use of the tool.

It's extremely cool to read about interview experiences like this. Thanks for the writeup, Ian. I hate combative interviews so much.

Re: What production means - blog dot lusis

phrenologist — Thu, 12 Jul 2012 16:50:24 -0000

Let's muddy the waters a bit with a different definition of production.

My group is Release Management. We manage all the software installed on thousands of servers. We have a very specific definition of production: a production server is one in the list of machines we do pushes to.

If we don't push to a machine, it's not in production. That's all we care about, full stop. There's a CapEx group that's supposed to put machines in this list and take them out as they are brought up or retired.

In practice, that's _generally_ (but not always!) the same list of servers as in your definition. The complication comes in with things like monitoring and alerting. There are different monitoring services used by different groups. Some of those systems do not consult our list of production servers before generating alerts. Thus, we deal constantly with reports of 'production' outages from other systems. We then have to check the servers against our 'official' list to determine if it's an actual production outage, or an alert from a retired host or something like that.

I find that managing the idea of 'production' is one of the most complicated issues we deal with on a daily basis. Everyone has a slightly different definition, depending on their particular worldview.

Great post as always, John.

@philiph

Re: HollenbackDotNet - Caltrain Clipper Failure

phrenologist — Thu, 12 Apr 2012 16:45:39 -0000

I travel between multiple zones and double tagging at any station works just fine. I've even double tagged in a zone outside of my monthly pass zones.

The double tagging just loads the pass on your card. It doesn't validate that you are actually traveling in the zones indicated on your monthly pass.

Note however that conductors do see this on their handheld scanners. If you have a zone 4-3 monthly pass and the conductor reads your clipper card in zone 2, the reader displays the discrepancy. It's up to the conductor to notice this however, and in my experience they only do about 50% of the time (particularly if they are very busy).

Thus you definitely still need to purchase proper paper zone upgrades if you are traveling beyond your monthly pass zones. You can definitely get caught for fare evasion if you aren't carrying a proper paper zone upgrade outside of your zones.

Re: Learn Code The Hard Way -- Books And Courses To Learn To Code

phrenologist — Mon, 09 Apr 2012 16:30:25 -0000

Hey Zed, guess you answered my question about Scala, Clojure, and Erlang :). Very cool analysis. I hadn't thought about bucketizing the TIOBE results.

Oh also here's a thread I started on the San Francisco Perl Mongers list, with some interesting thoughts on TIOBE: http://mail.pm.org/pipermai...

Re: HollenbackDotNet - Devops Perceptions

phrenologist — Wed, 28 Mar 2012 14:46:40 -0000

Thanks I fixed the name, sorry about that.

I agree his approach was kind of combative. However I also think it's important for the rest of us on the devops mailing list and in the community in general strive towards being more open and understanding in situations like this. I hate to see devops getting a bad name as a closed community that doesn't welcome (or at least tolerate) outside opinions.

Re: HollenbackDotNet - Devops Perceptions

phrenologist — Wed, 28 Mar 2012 14:45:49 -0000

Yeah I totally agree and I tried to call that out in my blog post as well.

Re: On the Importance of Lube

phrenologist — Mon, 12 Mar 2012 11:22:29 -0000

Serious question - is google+ viable? You and I are in vaguely similar tech circles (I think I follow you on g+ for example). However, I find there to be very little interesting posting happening on g+. My stream seems dominated by one or two prolific posters.

That's actually a big reason I'm a twitter adherent - although some people do tend to overpost on twitter, they can't really dominate. This is largely due to the 140 character post limit. On g+ on the other hand very popular people and constant posters take up a disproportionate amount of screen real estate.

Am I using google+ wrong?

Re: rsync: the swiss army chainsaw of backup utilities - Holistic Engineering

phrenologist — Sun, 04 Mar 2012 19:02:00 -0000

Have you thought about using rsnapshot? It uses rsync hardlinks too but wraps everything up nicely.

http://sysadvent.blogspot.c...

Re: HollenbackDotNet - Ge Defective Light Bulb Story

phrenologist — Thu, 01 Mar 2012 18:56:05 -0000

Yes - the socket is rated for 150w. I agree it seems weird but these sockets do exist.

Re: Devops a Wicked problem

phrenologist — Mon, 09 Jan 2012 20:30:42 -0000

I think "collaboration and resilience go hand in hand" is really the key to this whole thing. I see this on a daily basis in the software release process - teams that break down the silos produce resilient and robust solutions. Teams that go off and work on their own invariably produce inferior code which doesn't inter-operate correctly.

@philiph

Re: 2011-in-review - blog dot lusis

phrenologist — Mon, 02 Jan 2012 22:26:57 -0000

Well played, sir. My only wish is for twitter names for all the people you mentioned to make sure I'm following them all.

Re: Devops from a sysadmin perspective

phrenologist — Thu, 08 Dec 2011 11:04:18 -0000

Patrick I think you really hit the devops nail on the head in this article. I really agree that devops is all about collaboration.

Re: Sensu, a monitoring framework - PorterTech

phrenologist — Thu, 03 Nov 2011 11:08:53 -0000

Did you evaluate using ganglia as a replacement for nagios? I know ganglia has auto-discovery but I'm not sure how people are using it in the cloud.

Re: HOWTO: Random Number in Shell Script

phrenologist — Mon, 15 Aug 2011 13:06:35 -0000

Note that the bash random number generator uses the pretty crappy rand() algorithm to generate pseudorandom numbers. It also generates a signed 16 bit number, so the max value it generates is 32767.

I much prefer using perl to generate random numbers, as it uses a better random number algorithm and it creates arbitrarily large random numbers. For example:

$ perl -le 'print int rand(10000000000000000000)'
8492917251578582016

Re: San Jose Highway Diagram

phrenologist — Wed, 10 Aug 2011 20:42:06 -0000

You have outdone yourself, sir.

Re: Not All Navigation is the Same

phrenologist — Sun, 07 Aug 2011 22:20:10 -0000

It seems like Consumer Reports would be the ideal organization to do the sort of side by side GPS comparison you describe. However I don't think they've done that (although I am admittedly too lazy to google it).

Re: http://lusislog.blogspot.com/2010/09/distributions-and-dynamic-languages.html

phrenologist — Tue, 28 Dec 2010 16:54:28 -0000

What about the old idea of serving binaries over nfs, split by platform with symlinks and such? That's how we dealt with python libraries at a previous company. Kind of ugly, but it worked.

Re: Ping.fm / Uploaded Image from Kenneth Bowen

phrenologist — Fri, 02 Apr 2010 15:32:01 -0000

Yes! Your kids are awesome!

Re: x planes

phrenologist — Mon, 18 Jan 2010 14:35:20 -0000

That appears to be a different jet powered school bus than the one I photographed at the Watsonville Fly-In a few years back: Jet Powered School Bus

Re: Ping.fm / Uploaded Image from Phil Hollenback

phrenologist — Sun, 01 Nov 2009 18:24:45 -0000

Dang that is a cute kid.

Re: x planes

phrenologist — Sat, 24 Oct 2009 18:26:53 -0000

This plane lives at NASA-Ames in Mountain View, CA now. Here's a picture of it I took at the 1997 open house: http://www.flickr.com/photo...