<?xml version="1.0" encoding="utf-8"?>
<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Disqus - Latest Comments for cperciva</title><link>http://disqus.com/by/cperciva/</link><description></description><atom:link href="http://disqus.com/cperciva/comments.rss" rel="self"></atom:link><language>en</language><lastBuildDate>Wed, 03 Apr 2019 03:54:38 -0000</lastBuildDate><item><title>Re: FreeBSD/EC2: Community vs. Marketplace AMIs</title><link>http://www.daemonology.net/blog/2017-10-21-FreeBSD-EC2-community-vs-marketplace-AMIs.html#comment-4407514829</link><description>&lt;p&gt;In this case, the product/application is FreeBSD, which is open source...&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Wed, 03 Apr 2019 03:54:38 -0000</pubDate></item><item><title>Re: Designing resilient systems: Circuit Breakers or Retries? (Part 1)</title><link>http://localhost:4000/designing-resilient-systems-part-1#comment-4256480308</link><description>&lt;p&gt;In case nobody has mentioned it yet: "Load Balancer" is typoed in your diagrams as "Load Balanacer".  Thought I should mention it so you don't end up copying the same error into future network diagrams.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Tue, 25 Dec 2018 18:50:15 -0000</pubDate></item><item><title>Re: How to port your OS to EC2</title><link>http://www.daemonology.net/blog/2018-07-14-port-OS-to-EC2.html#comment-3988778429</link><description>&lt;p&gt;There are many things, including perl, python, golang, ruby, rust, awscli, and the amazon-ssm-agent, which can be useful in EC2.  They can also get in the way, e.g., if you want one version and the AMI comes with a different version preinstalled.&lt;/p&gt;&lt;p&gt;So I recommend providing a "bare bones" AMI.  Maybe provide a "everything except the kitchen sink pre-installed" AMI too, but once you have a bare bones AMI it's easy to create another AMI by adding to it.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Sat, 14 Jul 2018 16:00:47 -0000</pubDate></item><item><title>Re: FreeBSD/EC2 on C5 instances</title><link>http://www.daemonology.net/blog/2017-11-17-FreeBSD-EC2-C5-instances.html#comment-3884795978</link><description>&lt;p&gt;Not yet.  But 11.2 isn't out yet...&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Thu, 03 May 2018 16:52:57 -0000</pubDate></item><item><title>Re: Some thoughts on Spectre and Meltdown</title><link>http://www.daemonology.net/blog/2018-01-17-some-thoughts-on-spectre-and-meltdown.html#comment-3713142279</link><description>&lt;p&gt;Google says they notified Intel, AMD, and ARM in June 2017, yes.  No idea when or if they notified anyone else.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Wed, 17 Jan 2018 16:58:27 -0000</pubDate></item><item><title>Re: FreeBSD/EC2 on C5 instances</title><link>http://www.daemonology.net/blog/2017-11-17-FreeBSD-EC2-C5-instances.html#comment-3628440133</link><description>&lt;p&gt;Wow, I have no idea what happened while I was writing that sentence... thanks, fixed!&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Wed, 22 Nov 2017 12:38:23 -0000</pubDate></item><item><title>Re: IPv6 on FreeBSD/EC2</title><link>http://www.daemonology.net/blog/2017-01-26-IPv6-on-FreeBSD-EC2.html#comment-3395130764</link><description>&lt;p&gt;If you can ping out but you can't ping in, it's probably a security group problem.  The default security group doesn't allow IPv6.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Sat, 01 Jul 2017 14:12:55 -0000</pubDate></item><item><title>Re: A plan for open source software maintainers</title><link>http://www.daemonology.net/blog/2017-05-11-plan-for-foss-maintainers.html#comment-3301112367</link><description>&lt;p&gt;Wait, what?  My proposal is nothing at all like yet another freelancing job board.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Thu, 11 May 2017 17:53:07 -0000</pubDate></item><item><title>Re: A plan for open source software maintainers</title><link>http://www.daemonology.net/blog/2017-05-11-plan-for-foss-maintainers.html#comment-3299677633</link><description>&lt;p&gt;I had already decided that Connect would be a necessary part of this. :-)&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Thu, 11 May 2017 04:43:56 -0000</pubDate></item><item><title>Re: A plan for open source software maintainers</title><link>http://www.daemonology.net/blog/2017-05-11-plan-for-foss-maintainers.html#comment-3299498957</link><description>&lt;p&gt;Yes, github could provide this by adding subscription payment functionality and some logic for attributing sponsorship dollars to issues.  (I think that last part is important -- if a big company is sponsoring $1000/month and asks for something, I'm going to care more than if someone who isn't sponsoring asks for something.)&lt;/p&gt;&lt;p&gt;Or I suppose patreon could provide this by adding some sort of "requests box" functionality.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Thu, 11 May 2017 00:57:30 -0000</pubDate></item><item><title>Re: A plan for open source software maintainers</title><link>http://www.daemonology.net/blog/2017-05-11-plan-for-foss-maintainers.html#comment-3299483528</link><description>&lt;p&gt;I've seen lots of attempts along those lines; they tend to work well for getting particular features implemented, but fail when it comes to ongoing maintenance.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Thu, 11 May 2017 00:37:21 -0000</pubDate></item><item><title>Re: IPv6 on FreeBSD/EC2</title><link>http://www.daemonology.net/blog/2017-01-26-IPv6-on-FreeBSD-EC2.html#comment-3185203751</link><description>&lt;p&gt;Are you missing the accept_rtadv part in /etc/rc.conf ?  That's the only way I can imagine FreeBSD not getting the IPv6 routing.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Fri, 03 Mar 2017 16:49:07 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166938766</link><description>&lt;p&gt;You and your friend have watches which are synchronized to the second.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Tue, 21 Feb 2017 03:39:43 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166935100</link><description>&lt;p&gt;Correct!&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Tue, 21 Feb 2017 03:34:00 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166896328</link><description>&lt;p&gt;Correct!  And now I'm deleting your answer so that other people don't see it. :-)&lt;/p&gt;&lt;p&gt;(I'll bring it back later, don't worry.)&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Tue, 21 Feb 2017 02:35:05 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166870210</link><description>&lt;p&gt;7*2+10 and 5*3+9 are both equal to 24.  There are only 23 questions...&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Tue, 21 Feb 2017 01:55:13 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166751892</link><description>&lt;p&gt;Good.  But you can do better.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Mon, 20 Feb 2017 23:30:31 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166744316</link><description>&lt;p&gt;I was trying to convey that it isn't suspicious if he leaves early.  I'll clarify that in the question.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Mon, 20 Feb 2017 23:21:23 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166740115</link><description>&lt;p&gt;Your friend can leave at any time -- no need to worry about the proctors getting suspicious if he leaves after 89 minutes and 59 seconds.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Mon, 20 Feb 2017 23:16:38 -0000</pubDate></item><item><title>Re: Cheating on a string theory exam</title><link>http://www.daemonology.net/blog/2017-02-21-cheating-on-a-string-theory-exam.html#comment-3166728862</link><description>&lt;p&gt;Yes.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Mon, 20 Feb 2017 23:05:36 -0000</pubDate></item><item><title>Re: IPv6 on FreeBSD/EC2</title><link>http://www.daemonology.net/blog/2017-01-26-IPv6-on-FreeBSD-EC2.html#comment-3122663735</link><description>&lt;p&gt;I heard a rumour recently that VPC ELBs can have IPv6 turned on by AWS support -- hopefully that's an indication that it's going to be turned on for everybody in the near future.  But that was just an example; people may need other services which have yet to be IPv6-enabled.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Fri, 27 Jan 2017 03:12:39 -0000</pubDate></item><item><title>Re: EC2's most dangerous feature</title><link>http://www.daemonology.net/blog/2016-10-09-EC2s-most-dangerous-feature.html#comment-2942668559</link><description>&lt;p&gt;I always thought it was funny that Amazon says that you should not put anything sensitive in EC2 instance metadata, and then goes and puts AWS credentials into it.&lt;/p&gt;&lt;p&gt;Yes, there are ways to restrict access to the instance metadata, but it's very hard to do that while maintaining the functionality of IAM Roles.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Mon, 10 Oct 2016 05:53:57 -0000</pubDate></item><item><title>Re: EC2's most dangerous feature</title><link>http://www.daemonology.net/blog/2016-10-09-EC2s-most-dangerous-feature.html#comment-2941742766</link><description>&lt;p&gt;I never said that there weren't alternatives.  I'm just saying that IAM Roles is a very easy way to get yourself into trouble and people should be very careful when using it.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Sun, 09 Oct 2016 16:19:31 -0000</pubDate></item><item><title>Re: EC2's most dangerous feature</title><link>http://www.daemonology.net/blog/2016-10-09-EC2s-most-dangerous-feature.html#comment-2941740872</link><description>&lt;p&gt;Yes, you'd need to MITM attempts to access the metadata over HTTP... and while doing that, figure out which uid owns the socket making the request so that you can perform access control checks.  It's possible, but would involve a lot of groping inside the kernel -- far more than if Amazon supported loading IAM Role credentials from a location in the filesystem.  (And yes, this could be handled by having a filesystem which maps to the HTTP metadata service and blocking access to that service to everybody else, but presenting the data over XenStore would be much cleaner.)&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Sun, 09 Oct 2016 16:18:07 -0000</pubDate></item><item><title>Re: EC2's most dangerous feature</title><link>http://www.daemonology.net/blog/2016-10-09-EC2s-most-dangerous-feature.html#comment-2941735738</link><description>&lt;p&gt;IAM Roles provide time-limited credentials.  If you grab them out of the EC2 metadata and present them to utilities as "credentials on disk" then they'll be expected to be long-lived credentials and things will break when IAM Roles rolls the keys.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">cperciva</dc:creator><pubDate>Sun, 09 Oct 2016 16:14:16 -0000</pubDate></item></channel></rss>