Time Nick Message 12:59 pdurbin i just emailed a Mr. Thomas McGonagle ( http://www.meetup.com/Boston-DevOps-Meetup/members/20695341/ ), asking him if he'd like to join us here in #crimsonfu. I had exchanged emails with him about Puppet a while back. looks like he's on github too, and using vagrant: https://github.com/mcgonagletom/puppet_nist 13:03 pdurbin wow, this "Smashing the Stack in 2012: A brief tour of exploit mitigation techniques, and how to beat them" talk tonight at 7pm sounds great: https://lists.hcs.harvard.edu/pipermail/hcs-announce/2012-April/000490.html 13:04 pdurbin the speaker also gave a talk called "Breaking out of KVM": http://blog.nelhage.com/2011/08/breaking-out-of-kvm/ 13:06 SEJeff_work Oh those look very fun 13:06 SEJeff_work totally envious 13:08 pdurbin the problem is that i'm afflicted with children ;) who are far to young to attend such talks with me, and who need to be picked up from day care :) 13:15 SEJeff_work Ha 13:15 SEJeff_work The smashing the stack talk is clearly a play on the *classic* article from Phrack 13:15 SEJeff_work http://insecure.org/stf/smashstack.html 13:16 SEJeff_work That was a long long time ago 13:19 pdurbin i feel like i should mention why i hestiated to put /var/named under version control (git). I was listening to a git talk by Randal Schwartz and he says (on slide 4) that git is "not for... tracking file permissions and ownership" http://www.slideshare.net/RandalSchwartz/introduction-to-git-11451326 . he says for stuff in /etc, he uses RCS instead 13:20 pdurbin so i'm unclear if using git the way i am is a problem. . . i don't think it is. . . 13:27 pdurbin anyway, it seems to be working fine. the way i expect. i'll report back if i encounter any problems 13:41 SEJeff_work pdurbin, git does not preserve file permissions or ownership. That is correct 13:41 SEJeff_work Randall is a bit odd. He is in #salt often as RandalSchwartz 13:42 SEJeff_work Really seasoned admin, but a little too unix beardy for my tastes. We've went out a few times for drinks as he works with some of my friends here in Los Angeles 13:44 pdurbin i met him *very* briefly, as i mentioned previously: http://irclog.perlgeek.de/crimsonfu/2012-04-12#i_5432840 13:45 shuff hey folks 13:45 shuff it's good to be back :) 13:46 pdurbin hey shuff. anyway, git is becoming a "go to" tool for me. if it does what i need i'm not going to go learn rcs 13:50 pdurbin the way i've got /var/named in git is pretty much the way cobbler does it 13:51 SEJeff_work pdurbin, If you don't care about commit messages, you could also use rsnapshot 13:51 SEJeff_work Which is really nice software for what it does 13:51 pdurbin yeah, we use rsnapshot, but i *do* want commit messages, even if they're all as the same git author 13:52 pdurbin (i'm referring to /var/lib/cobbler/.git . the git author is always "API update") 13:53 pdurbin anyway, i should have put /var/named in git months ago. i'm way less grumpy about tweaking DNS now :) 13:55 pdurbin imagine that. a grumpy sysadmin 13:55 agoddard pdurbin: why not use config. management for DNS? 13:55 pdurbin agoddard: would love to. do you? 13:56 agoddard node[:dns][:zones].each do |zone| < yup :D 13:56 pdurbin is it on github? 13:57 agoddard chef already knows the fqdn of all the nodes it manages, and their IPs, so I wrote a recipe that dumps that into the DNS config. Then additional records (CNAMES etc) are just k->v pairs in attributes 13:57 agoddard https://github.com/mbl-cli/cli-legacy-cookbooks/blob/master/maradns/recipes/primary.rb#L34 13:58 agoddard (disclaimer, I hate maradns, swapping it out when we have time in an iteration) 13:58 agoddard https://github.com/mbl-cli/cli-legacy-cookbooks/blob/master/maradns/templates/default/zone.erb 14:01 * pdurbin looks at http://en.wikipedia.org/wiki/MaraDNS 14:01 agoddard (the k-> pairing is stored in git too, but it's private ) 14:01 pdurbin agoddard: thanks. very interesting 14:02 pdurbin agoddard: swapping out maradns for what? BIND? 14:03 agoddard pdurbin: probably BIND. Haven't thought about it much, but will be trivial to take this template and make it work for any other DNS server 14:07 agoddard pdurbin: how heavily do you guys use puppet? 14:07 pdurbin pretty heavily! despite my playing with salt, i love puppet 14:08 agoddard schweet 14:09 shuff there are a bunch of modules on the forge tagged as "dns"; i haven't yet looked through all of them 14:10 pdurbin oh hey! "@philipdurbin I'm not going to discuss that 140 chars at a time. Please contact me by email. My address is all over the web. :)" -- http://twitter.com/merlyn/status/195145984257429504 14:11 SEJeff_work pdurbin, You know... 14:11 SEJeff_work We asked Mike for that feature eons ago 14:11 SEJeff_work That specific feature 14:12 pdurbin who's mike? 14:12 SEJeff_work Maybe 3 years ago we asked him to back it with git, but it is slow, you need fast storage for that feature 14:12 SEJeff_work Michael Dehaan is the guy who wrote cobbler. Used to work for redhat 14:12 SEJeff_work As one of his bigger users at the time, we worked quite closely with him 14:13 pdurbin i don't understand. . . our cobbler is in git. i just had to flip some bit in /etc 14:13 SEJeff_work http://hastebin.com/gubudehasu.vhdl 14:13 SEJeff_work ^^ /etc/cobbler/settings 14:14 pdurbin "scm_track_enabled: 1" 14:15 SEJeff_work We asked for that feature in specific as we have 3 admins in 2 offices making changes and it was getting a bit harder to see what was going on. Longer term... I wrote this thing that talks to cobblerd's xmlrpcapi, and puts every system record into a database daily and then emails out a pretty unified diff of every added, modified, or removed system. I might be able to convince work to let me oss that 14:15 pdurbin ok, so all your saying is that you requested that feature. thanks! i love it! 14:15 SEJeff_work yes 14:16 SEJeff_work Mike is one of those rare project maintainers who is really good about being on IRC most of the time 14:16 SEJeff_work So you can bounce ideas off him which lead to results very quickly 14:17 pdurbin sounds like thomas hatch, the lead developer of salt 14:35 SEJeff_work Pretty much, yes 14:35 agoddard favorite (re)tweet of the week: https://twitter.com/#!/miller_joe/statuses/176815604869758977 14:37 pdurbin agoddard: heh. yeah, i saw that. awesome :) 15:25 pdurbin just linked a user to http://wiki.greptilian.com/mysql/mysql_secure_installation/ 15:34 pdurbin i guess i'm going to that "smashing the stack" talk tonight, if anyone wants to meet me there 15:36 pdurbin i can't wait to have more to worry about 15:48 SEJeff_work Have more children then. Should solve the problem 15:51 pdurbin -1 15:52 shuff having more children protects you against buffer overrun attacks? i think you lost me there :) 15:54 ironcamel children++ 15:54 ironcamel pdurbin: what is "smashing the stack" ? 15:55 SEJeff_work ironcamel, <pdurbin> wow, this "Smashing the Stack in 2012: A brief tour of exploit mitigation techniques, and how to beat them" talk tonight at 7pm sounds great: https://lists.hcs.harvard.edu/pipermail/hcs-announce/2012-April/000490.html 15:57 ironcamel ah 16:06 pdurbin ironcamel: too bad you aren't local 16:43 SEJeff_work Any taskwarrior fans here? 16:52 pdurbin shuff is: http://irclog.perlgeek.de/crimsonfu/2012-04-12#i_5432337 17:02 ironcamel A friend and I just released App::Notes https://metacpan.org/module/notes 17:03 ironcamel the neat thing is that it is backed by git, so you can create a gist for example, then run notes init uri_of_your_gist 17:03 ironcamel and you get a free web interface to your notes 17:03 pdurbin sounds like ikiwiki 17:03 ironcamel a new note is just a new file in your gist 17:04 ironcamel notes add and notes edit open up vim for you for you to edit your note 17:04 ironcamel unless you are piping in stdin, in which case it just uses the content of stdin 17:05 pdurbin i'll have to play with it 17:05 ironcamel pdurbin: sudo cpanm notes 17:05 ironcamel to install it 17:05 ironcamel or sudo cpan App::Notes 17:05 pdurbin ok. for ikiwiki, http://wiki.greptilian.com is backed by http://git.greptilian.com/?p=wiki.git 17:08 ironcamel neat. the cool thing about apps backed by git is you get history for free. 17:09 pdurbin absolutely. forget mysql as a backend. just use git 17:09 ironcamel make a mistake, just git reset blahblahblah 17:11 pdurbin yep 18:07 pdurbin ironcamel: have a sample git repo? for App::Notes? 20:06 pdurbin looking at http://git.fedorahosted.org/git/?p=linux-pam.git;a=blob;f=modules/pam_access/access.conf;hb=HEAD 20:07 SEJeff_work pdurbin, If you use rhel, don't use pam_access 20:08 SEJeff_work If you use RHEL 6.x or 5.7+, you have sssd 20:09 SEJeff_work And sssd has the simple auth provider, which replaces pam_access. Ironically, I asked sgallagh, the sssd project lead, to add group support to it so we could phase out pam_access entirey: http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/6/html/Deployment_Guide/config-sssd-domain-access.html 20:11 SEJeff_work pdurbin, you do a lot of similar stuff that I do. Maybe if we're ever in the same area, we can pick eachother's brains 20:12 SEJeff_work http://linux.die.net/man/5/sssd-simple 20:17 shuff SEJeff_work: wait, what? we're using pam_access on rhel6 and ubuntu lucid over here, and it appears to work just fine 20:17 shuff but i'll take a look at that simple auth 20:17 SEJeff_work shuff, Oh it does work just fine, it is just a bit annoying 20:17 SEJeff_work or I <3 sssd more overall 20:17 shuff pdurbin: i can toss you the puppet module that we use to manage pam_access if you so desire 20:18 shuff it will be on the forge eventually 20:22 pdurbin shuff: i have some pam_access config that came from Pax, originally, I think. i dunno 20:23 pdurbin SEJeff_work: yes, beers some day. and i'll look at sssssssd. thanks 20:23 SEJeff_work If you use authconfig or anaconda to configure ldap and or krb in RHEL, you already use it 20:24 pdurbin this is the thing when you start a new job. you have to get up to speed with how things are done there 20:25 pdurbin which is always slightly different than the last shop 20:27 pdurbin shuff: did we use pam_access at hmdc? i guess whorka would know 20:28 pdurbin this is what i was looking for, by the way: http://git.fedorahosted.org/git/?p=linux-pam.git;a=blob;f=modules/pam_access/access.conf;hb=HEAD#l45 20:29 pdurbin "To avoid problems with accounts, which have the same name as a group..." 20:29 pdurbin just wanted to make sure i understood the syntax 20:34 Pax hey folks! I was just looking over the red hat docs for sssd simple for access control, does it do network based control as well? Or just user /group controls? 20:34 SEJeff_work Pax, What do you mean network access control? 20:35 SEJeff_work It supports standard netgroups similar to normal ldap or nis 20:35 SEJeff_work but if you use their IPA server, it supports their HPAC, which is more flexible than netgroups. 20:36 Pax so, part of what I initially liked about pam_access was that it had the sort of tcpwrappers like ability to allow user foo from network 1.2.3.0/24 and only allow them access from that net 20:36 Pax so like +: foo : 1.2.3.0/24 20:37 SEJeff_work Oh touche! sssd doesn't do that, but you could do it with netgroups / hbac 20:37 SEJeff_work Not near as easy as pam_access for that however 20:39 Pax gotcha! I wouldn't have been surprised if sssd had been able to do it though, I feel like i've only touched the surface of it's awesomeness 20:39 Pax er and by gotcha I mean "I follow what your saying" 20:40 Pax geez it really is the end of the day huh? 20:42 pdurbin shuff: is there an easy way to get a border around a PHP Markdown Extra table? http://michelf.com/projects/php-markdown/extra/#table just in markdown. no css, i mean 20:42 shuff uhhhhhhhhhhh 20:42 shuff i have no idea whatsoever :( 20:42 pdurbin :) that's ok. i need to run anyway 20:42 shuff i have avoided having anything to do with any of the markdown table implementations 20:44 pdurbin i've changed our markdown processor 3 times already. whatever jekyll uses, then Text::Markdown, now discount 20:44 pdurbin i think i'm done. really 20:44 SEJeff_work Pax, yup, that entire team (redhat's IPA / sssd) team is full of smart guys. Worked with them (on sssd) since pre-1.0. It is fantastic software when compared to pam_ccreds or nscd's bag o fail 20:45 pdurbin the only thing discount doesn't have is the github-flavored markdown "don't squish newlines together" feature 20:45 SEJeff_work What does gruber use? 20:45 shuff he disdains the vile table 20:45 Pax SEJeff_work re: nscd OMG truer words were never spoken 20:45 pdurbin "The biggest difference that GFM introduces is in the handling of linebreaks" -- http://github.github.com/github-flavored-markdown/ 20:46 SEJeff_work Pax, infinite negative caching what! 20:46 pdurbin i basically want discount to implement github-flavored markdown 20:46 SEJeff_work pdurbin, You know gfm is in a github repo, right? 20:46 SEJeff_work You can just steal it and use it :) 20:46 pdurbin can ikiwiki use it? 20:46 SEJeff_work https://github.com/github/github-flavored-markdown 20:47 SEJeff_work it is software, you can make software do anything. They probably wouldn't integrate without some love, but it might be possible. 20:47 pdurbin amen. ok. gotta go