IRC log of #zope for Thursday, 2014-09-18

*** stevegt_ has joined #zope00:09
*** alecm has quit IRC00:27
*** alecm has joined #zope00:28
*** alecm has quit IRC00:43
*** supton has quit IRC00:45
*** alecm has joined #zope00:46
*** acsr has quit IRC01:05
*** menesis has quit IRC01:12
*** acsr has joined #zope01:27
*** acsr1 has joined #zope01:40
*** acsr has quit IRC01:40
*** daMaestro has quit IRC01:51
*** agroszer has quit IRC01:55
*** acsr has joined #zope01:58
*** acsr1 has quit IRC01:58
*** acsr1 has joined #zope02:06
*** acsr has quit IRC02:06
*** acsr has joined #zope02:09
*** acsr1 has quit IRC02:09
*** m8 has quit IRC02:16
*** kosh has joined #zope02:38
*** acsr1 has joined #zope02:51
*** acsr has quit IRC02:51
*** acsr has joined #zope02:55
*** acsr1 has quit IRC02:58
*** do3cc has quit IRC03:13
*** acsr1 has joined #zope03:15
*** acsr has quit IRC03:15
*** do3cc has joined #zope03:26
*** sm has quit IRC03:31
*** sm has joined #zope03:33
*** sm has quit IRC03:38
*** sm has joined #zope03:41
*** acsr1 has quit IRC03:45
*** acsr has joined #zope03:45
*** acsr has quit IRC03:53
*** acsr has joined #zope04:01
*** alecm has quit IRC04:17
*** alecm has joined #zope04:18
*** acsr1 has joined #zope04:42
*** acsr has quit IRC04:42
*** acsr1 has quit IRC05:09
*** alecm has quit IRC05:46
*** __mac__ has joined #zope06:58
*** __mac__ has quit IRC07:21
*** tisto has joined #zope08:00
*** alecm has joined #zope08:02
*** PeterZ1 has joined #zope08:33
*** kosh has quit IRC08:37
*** PeterZ1 has left #zope08:46
*** alecm has quit IRC09:02
*** avoinea has joined #zope09:11
*** alecm has joined #zope09:11
*** alecm has quit IRC09:24
*** Spanktar has joined #zope09:41
jakkeI'm still stuck here, anyone an idea why document_view appears in the undo log every now and then?10:11
jakkewhen the document_view fails to resolve the portal id and username, it also messes up portlets and corrupting data10:11
betabugmaybe #plone is a better place to ask this?10:13
betabugthere are no portlets in zope10:13
jakkeyes and no10:13
jakkeI don't think it's purely portlets related10:13
jakkeas far as I can tell now, it seems to be related to the ZPublisher10:14
betabugthere is also no portal id, so I don't know what you mean with "fails to resolve the portal id"10:14
betabugthinking of it, dunno what your document_view is either10:14
jakkeok ok, point made :D10:15
jakkeI'm in zope code, hence I popped the question here first :)10:15
betabugif you suspect it's a zope problem, you'd need to find a test case without all the plone stuff10:16
jakkebtw betabug, do you ever sleep, or do you happen to be in the european timezone?10:16
betabugOTOH, #plone is probably a lot more active than #zope10:16
jakkethat's my main problem, I haven't found a way to reproduce the problem10:16
betabugI'm in europe10:16
jakkemakes sense now :)10:18
*** Gogo|tty has quit IRC10:39
*** River-Rat has quit IRC10:43
*** River-Rat has joined #zope10:44
*** dixond_ has quit IRC10:50
*** dixond has joined #zope10:51
*** regebro has quit IRC10:53
*** regebro has joined #zope10:56
*** Gogo|tty has joined #zope11:00
*** menesis has joined #zope11:48
*** agroszer has joined #zope13:00
*** menesis has quit IRC13:52
*** jakke1 has joined #zope13:56
*** jakke has quit IRC13:59
regebroPing J1m_ ! Just wanted to see if you saw my email on Buildout and sudo/setuid per part?14:11
*** menesis has joined #zope14:31
*** fredvd has joined #zope14:32
*** J1m_ has quit IRC14:39
*** aclark has joined #zope14:55
*** kosh has joined #zope15:38
*** J1m_ has joined #zope16:18
*** do3cc has quit IRC16:30
*** do3cc has joined #zope16:40
J1m_regebro, I saw your email and your tweet. I'm sorry I haven't had time to respond.16:43
J1m_I've been traveling a lot the last few days.16:43
regebroJ1m_: No hurry, I just wanted to make sure you saw it. Take your time.16:43
J1m_we run buildout as root all the time.16:43
J1m_but we launch it from higher-level processes that run as root, so no need for setuid.16:44
J1m_we also split our buildouts into "build" and "deploy" configurations.16:44
J1m_This is what we run it from:
J1m_Sadly, I made the mistake of tying this too closely to our monitoring framework so it's not reusable. I'm working on fixing this, but ...16:46
J1m_and then there's Docker...16:46
*** Feneric has joined #zope16:57
*** agroszer has quit IRC17:03
mcdonci'm struggling with using ansible myself.. afaict from lurking in #ansible there's some sort of belief that you can just run the same playbook against an already-provisioned host and you can write the playbook in such a way that it always does the right thing.. pretty sure this is just wishful thinking17:05
J1m_mcdonc, I'm really pretty happy with zkdeployment.17:11
mcdoncyeah, i like the writeup17:11
J1m_It let's us have a very high-level model of our systems.17:11
J1m_docker could make it a lot simpler.17:11
J1m_at least the deployment parts.17:12
J1m_we're already using it to deploy a few docker containers.17:12
mcdoncwould you just build the containers locally and push them out?17:12
J1m_I'd build images locally, push them to a private repo (S3 backed) and zkdeployment to deploy containers from them.17:13
J1m_Ideally, the images would be used for CI.  We're only using docker for some semi-legacy FTP servers for now.17:14
mcdoncseems to make sense; that way you always "provision", you never really "configuration manage"17:16
mcdoncevery creation of a container is a new provisioning17:16
J1m_yeah, although if you use docker as intended, there's still a bit of wireing in production.17:16
J1m_although we use ZooKeeper to do most of that.17:17
mcdoncthere's still the problem of managing the system that contains all the state (a database) where you cant throw it out and start over cuz its got data17:18
mcdoncwhere "problem" == need to update-in-place i think17:19
*** alecm has joined #zope17:20
*** alecm has joined #zope17:20
J1m_A container can use data outside itself.17:20
J1m_so, four or ftp servers and, the data are mounted into the container from the host.17:21
J1m_We can replace the software by installing a new image without affecting the data.17:21
J1m_The data are on ebs volumes, so we can replace the AWS instances as well without losing data.17:21
mcdoncso i guess you could mount an ebs volume to do same for e.g. postgres or zodb data?  i've never used aws.17:22
mcdoncpretty fly17:23
J1m_Now the "standard" ebs volumes are SSDs.17:25
regebroJ1m_: The whole setuid question is prompted by me trying to install a Openstack development environment and failing.17:25
J1m_why do you need root in development?17:26
regebroTo solve my actual problem, docker may be an option.17:26
regebroJ1m_: To install system packages.17:26
regebroSee also
regebroSystem packages also, and annoyingly need to be installed remotely on the open stack servers, but I doubt buildout would be the solution to that.17:27
regebroJ1m_: In any case, this is on a rather theoretical level at the moment.17:27
regebroJ1m_: One solution that does not add the feature but to add some sort of hooks for an extension between the steps.17:29
J1m_Buildout shouldn't call itself a build system.17:30
regebroBut if you don't want that, then I'll just skip it. I'm not 100% sure I'm going to end up using it anyway, so it would definitely be in a branch until I end up actually using it. :-)17:30
J1m_I never call it that.17:30
J1m_I call it an assembly system :) But no one ever understand that. :)17:31
regebroJ1m_: Good, yeah, I think that's wrong too.17:31
J1m_I don't have time to read your blog post atm. But I will later. Looks interesting.17:32
regebroThe buildout website needs some love, that's for sure. But I don't see any free time for me this year.17:32
regebroSure, no pb.17:32
*** agroszer has joined #zope17:36
mcdoncmy main problem right now is trying to convince other team members that update-in-place is nondeterministic17:37
mcdoncnot a problem i can solve with tools, unfortunately17:37
J1m_mcdonc, at least buildout makes update in place fairly sane.17:37
J1m_because buildout cleans up after itself.17:38
mcdoncyeah; we actually use it (driven by ansible)17:38
mcdoncits all the nonbuildout things that are problematic17:38
J1m_It's crazy how many established configuration systems don't do that.17:38
J1m_I'm fairly convinced that most system management culture was established when managing minicomputers for scientists.17:40
mcdoncthey can but the temptation to just write "cp /srcstuff /deststuff" or other things that dont take existing state into account is strong17:40
mcdoncyou saw stuff like this at usgs?17:40
J1m_I see this today when I talk to systems people or study system management tools.17:42
J1m_for example the common notion of convergence of system configuration makes no sense when deploying services automatically.17:42
J1m_It might make sense as part of a process of finely hand crafting machines.17:43
mcdoncafaict the allure of fabric/ansible/salt is that people believe they understand how to create imperative things and that they can understand them after they're created, but its much harder to think about creating something entirely declarative (which is why C/python/ruby is more popular than haskell, i guess)17:43
J1m_And I have sympathy for that PoV if you run the configuration once.17:44
mcdoncthey'll write cp /foo /bar instead of creating a recipe that makes sure foostuff-is-in-par17:44
J1m_The real power of recipies is that they know how to uninstall.17:46
J1m_Otherwise it's just a matter of abstraction.17:47
mcdoncyeah.. and the logic to drive whether something should install or uninstall-them-install is put into a nice little box17:47
J1m_If you didn't use something like buildout, you'd want to at least have a library with functions for common tasks.17:47
mcdoncand all the conditionals are driven up there17:47
mcdonci dont have much faith that a random team could create something that updates in place properly tho... the business pressures are such that taking 2 hrs to create the recipe that does that would be considered a waste of time17:50
mcdoncwhich is why i think that the only hope for j. random team is to always rebuild17:50
*** Arfrever has quit IRC18:04
*** Arfrever has joined #zope18:28
*** tisto has quit IRC19:24
*** PeterZ1 has joined #zope19:42
*** PeterZ1 has quit IRC19:42
*** regebro has quit IRC20:00
*** regebro has joined #zope20:00
*** m8 has joined #zope20:03
regebroThe uninstall part of having something declarative is something I didn't even mention in my blog post. :-P20:03
* regebro waves20:03
*** regebro has quit IRC20:03
*** kiorky has quit IRC20:09
*** kosh has quit IRC20:10
*** kiorky has joined #zope20:11
*** J1m_ has quit IRC20:17
*** J1m_ has joined #zope20:19
*** Spanktar has quit IRC20:28
*** menesis has quit IRC21:06
*** menesis1 has joined #zope21:06
*** fredvd has quit IRC21:07
*** menesis1 has quit IRC21:12
*** J1m_ has quit IRC21:16
*** J1m_ has joined #zope21:22
*** J1m_ has quit IRC21:26
*** J1m_ has joined #zope21:28
*** jham_ has quit IRC21:48
*** rbanffy has quit IRC21:56
*** jham has joined #zope22:01
*** avoinea has quit IRC22:03
*** jhammons has joined #zope22:13
*** kiorky has quit IRC22:16
*** Pumukel has joined #zope22:37
*** kiorky has joined #zope23:04

Generated by 2.15.1 by Marius Gedminas - find it at!