#archlinux-ports | Logs for 2017-06-20

Back
[00:09:28] -!- tyzoid has quit [Ping timeout: 258 seconds]
[00:21:47] -!- tyzoid has joined #archlinux-ports
[02:15:01] -!- isacdaavid has quit [Remote host closed the connection]
[02:15:01] -!- piernov_ has joined #archlinux-ports
[02:15:02] -!- piernov has quit [Remote host closed the connection]
[08:19:28] -!- titus_livius has joined #archlinux-ports
[08:19:46] <deep42thought> did I miss something? my logger died :-(
[08:23:06] <phillid> Not too much
[08:23:18] <phillid> t​yzoid has gone to bed
[08:23:32] <phillid> Also introduced me to the forum
[08:24:46] <deep42thought> ok, thanks for the update
[08:39:55] <deep42thought> tyzoid: my support set up your DKIM key (it seems, their web frontend does not yet allow subdomains starting with "_")
[10:19:10] -!- isacdaavid has quit [Quit: isacdaavid]
[10:39:30] -!- tyzoid has quit [Ping timeout: 240 seconds]
[10:50:48] -!- dmakeyev has joined #archlinux-ports
[10:51:57] -!- tyzoid has joined #archlinux-ports
[11:32:38] -!- Faalagorn has joined #archlinux-ports
[12:12:53] piernov_ is now known as piernov
[12:25:02] -!- dmakeyev has quit [Ping timeout: 240 seconds]
[12:28:25] -!- Faalagorn has quit [Quit: Leaving.]
[12:49:21] -!- dmakeyev_ has joined #archlinux-ports
[12:58:07] -!- dmakeyev_ has quit [Ping timeout: 255 seconds]
[13:54:10] -!- tyzoid has quit [Ping timeout: 258 seconds]
[14:07:23] -!- tyzoid has joined #archlinux-ports
[16:16:56] -!- deep42thought has quit [Quit: Leaving.]
[16:25:57] -!- tyler1 has joined #archlinux-ports
[16:26:38] -!- tyzoid has quit [Disconnected by services]
[16:26:43] tyler1 is now known as tyzoid
[16:27:03] -!- tyzoid_ has joined #archlinux-ports
[16:27:07] <tyzoid> lol
[16:28:48] <tyzoid> missed deepthought by 10 minutes
[16:37:29] <tyzoid> Thanks deep42thought: With the DKIM addition, I now get 10/10 on mail-tester
[16:37:31] <tyzoid> https://www.mail-tester.com
[16:53:36] <rewbycraft> Nice
[17:24:51] -!- deep42thought has joined #archlinux-ports
[17:27:57] <deep42thought> tyzoid: nice
[17:32:57] <deep42thought> btw: I moved most of the testing packages to stable (and removed a pile of bugs from my scripts on the way)
[17:33:08] <tyzoid> sweet
[17:33:17] <tyzoid> I'll do another installation test when I have a break from work
[17:34:17] <deep42thought> ok
[17:34:45] <deep42thought> the packages currently in "testing" are ones, that have been moved there from "staging" (this is done automatically)
[17:35:30] <tyzoid> alright, sweet
[17:35:34] <tyzoid> I'll do a test of that as well
[17:35:50] <tyzoid> Have we thought any more about testing methodology?
[17:36:21] <tyzoid> Apart from the installation tests, which just make sure nothing has horribly broken, we should have a document or something that outlines what testing we can do/should implement
[17:37:29] <deep42thought> the question is how much backlog we find acceptable
[17:39:08] <deep42thought> I'll create some discussion thread on the forum
[17:42:02] <tyzoid> Not sure what you mean by backlog
[17:42:42] <tyzoid> Once we start automating our tests, there shouldn't be a backlog of packages awaiting testing
[17:42:44] <tyzoid> if that's what you mean
[17:42:58] <deep42thought> that's what I meant
[17:43:19] <deep42thought> I'd like at least _some_ manual testing
[17:43:32] <deep42thought> but this would imply a larger delay
[17:43:49] <tyzoid> Even doing installation tests, we could automate
[17:44:38] <tyzoid> just fire up a virtualmachine with an install script
[17:45:05] <deep42thought> yeah, but for _really_ important packages we should do some package-specific tests (kernel: does it boot?, pacman: does it install packages?, ...)
[17:45:11] <tyzoid> That's what I mean
[17:45:17] <tyzoid> the script would do that upon installation
[17:45:19] <deep42thought> and I'm not sure if we can automate all this
[17:45:24] <tyzoid> I don't see why not
[17:45:28] <tyzoid> Here's my thinking
[17:45:38] <tyzoid> when the iso boots up, we have control over the files there
[17:45:54] <tyzoid> We can place an install-test script inside of the iso
[17:46:06] <tyzoid> that would format /dev/sda1 to ext4, and mount it as /mnt
[17:46:20] <tyzoid> then, we pacstrap it, and include a test package as well
[17:46:36] <tyzoid> then arch-chroot is called, loading the test script inside of the chrooted environment
[17:46:38] <tyzoid> instead of bash
[17:46:47] <tyzoid> which finishes the installation + the configuration of grub
[17:46:54] <tyzoid> then it exits, and reboots the vm
[17:47:10] <tyzoid> It would complete upon pinging a server with logs+status of success
[17:47:35] <deep42thought> hmm, sounds like a huge effort
[17:47:35] <tyzoid> what are your thoughts deep42thought?
[17:47:43] <tyzoid> I don't really think so
[17:47:44] <deep42thought> we can't do this for all packages
[17:47:47] <deep42thought> I think
[17:47:47] <tyzoid> Right
[17:47:55] <tyzoid> but that's for the installation + tests of core functionality
[17:47:58] <deep42thought> for important ones, that should work, though
[17:47:58] <tyzoid> like the archiso
[17:48:03] <deep42thought> yeah
[17:48:03] <tyzoid> plus it tests that linux works
[17:48:05] <tyzoid> grub works
[17:48:07] <tyzoid> pacman works
[17:48:08] <deep42thought> yes
[17:48:12] <tyzoid> and internet works
[17:48:32] <tyzoid> That's basically what I'm checking when I do the installation test
[17:48:37] <tyzoid> it's just basically an automation of that
[17:48:39] <rewbycraft> How do you wanna do the vm stuff?
[17:48:45] <tyzoid> virtualbox in 32bit mode
[17:48:50] <rewbycraft> Fair enough
[17:48:58] <tyzoid> It can be done using vboxmanage commands
[17:49:04] <rewbycraft> Hmh.
[17:49:14] <deep42thought> we could keep the iso fixed and provide the test-script by other means
[17:49:26] <rewbycraft> Maybe add a second cd-drive
[17:49:31] <deep42thought> this would remove the trouble of recompiling the iso each and every time
[17:49:34] <deep42thought> yeah
[17:49:38] <tyzoid> possibly shared folders?
[17:49:45] <rewbycraft> Nah, that needs vbox addons installed
[17:49:48] <deep42thought> or just pull some git repo
[17:49:52] <rewbycraft> Or that
[17:50:00] <tyzoid> but pulling a git repo requires a script to run on boot
[17:50:05] <deep42thought> that way we can also easily modify the tests
[17:50:07] <tyzoid> because it would need to be automated
[17:50:12] <deep42thought> yes
[17:50:23] <tyzoid> so we would need xdotool or something to run those commands
[17:50:29] <tyzoid> and it becomes slightly more complex
[17:50:31] <deep42thought> so we would need _one_ special iso
[17:50:46] <deep42thought> s/so/or/
[17:51:24] <tyzoid> I think the one special iso might work
[17:51:31] <tyzoid> but then it wouldn't be a test of the iso builds
[17:51:39] <tyzoid> which we'd have to make a separate test for in that case
[17:52:09] <deep42thought> I think, we can skip the iso test if the modification is minimal
[17:52:39] <deep42thought> e.g. only one additional activated systemd-unit + executable w.r.t. releng
[17:53:04] <tyzoid> I'm thinking back to the recent glibc bug
[17:53:28] <deep42thought> yes
[17:53:40] <tyzoid> I'd really like to have those isos tested prior to publishing themm
[17:53:47] <tyzoid> and automating that process makes it easier on us
[17:54:08] <tyzoid> Because the arch mainline project has many volunteers managing the packages
[17:54:14] <tyzoid> we have what, 3? 4?
[17:54:25] <tyzoid> We can't handle that volume without automating everything we can
[17:54:31] <deep42thought> yes
[17:54:42] <tyzoid> sorry if I'm preaching to the choir here
[17:55:11] <tyzoid> but I find the nonchalant attitude toward testing even on mainline to be a bit disturbing
[17:56:07] <deep42thought> once we have an automated test procedure, it should be simple to make more complex tests, right?
[17:56:19] <tyzoid> depends on how robust that infrastructure is
[17:56:25] <deep42thought> hmm
[17:56:30] <tyzoid> and what we're trying to test
[17:56:41] <tyzoid> testing installation requires a different set of infrastructure to testing packages
[17:56:51] <tyzoid> because on package testing, we can have a vagrant machine already available
[17:57:09] <tyzoid> just vagrant up, pacman -S archlinu32-keyring && pacman -Syu
[17:57:12] <tyzoid> then run the package test
[17:57:13] <deep42thought> installation testing can test both
[17:57:24] <deep42thought> can't it?
[17:57:27] <deep42thought> btw: I'll have to leave again in a few minutes
[17:57:29] <tyzoid> It can
[17:57:31] <tyzoid> but it'd take more time
[17:57:38] <tyzoid> and if we have 2000+ packages awaiting testing
[17:57:44] <tyzoid> that's not really feasable
[17:57:53] <tyzoid> unless we run a bunch in parallel
[17:58:02] <tyzoid> and that's extra compute power + unnecessary bandwidth
[17:58:11] <tyzoid> np deep42thought
[18:00:12] <deep42thought> hmm, the time might be an important issue ...
[18:00:35] <tyzoid> That why I was thinking an already installed vagrant vm might be more use here
[18:00:49] <tyzoid> plus we can automate building of that one vm via the automated install procedure periodically
[18:01:03] <deep42thought> good idea
[18:01:05] <tyzoid> it's only a few additional steps to make it a vagrant compatable box
[18:01:26] <deep42thought> ok
[18:01:34] <deep42thought> I'll leave now, cu later
[18:01:41] <tyzoid> Okay, see ya
[18:01:45] <deep42thought> btw: there is now a brainstorming thread on the forum
[18:01:48] -!- deep42thought has quit [Quit: Leaving.]
[18:16:05] -!- Faalagorn has joined #archlinux-ports
[18:20:52] -!- Faalagorn has quit [Ping timeout: 260 seconds]
[18:39:41] <tyzoid> so rewbycraft: I posted my chat over in #archlinux about this to the forum
[18:39:43] <tyzoid> https://bbs.archlinux32.org
[18:40:02] <tyzoid> They said that archlinux-arm might have some automated systems we could look at
[18:40:23] <rewbycraft> Interesting. Maybe worth poking them about it
[18:41:44] -!- tyzoid_ has quit [Ping timeout: 246 seconds]
[18:55:26] -!- tyzoid_ has joined #archlinux-ports
[19:45:09] <tyzoid> So rewbycraft: I was briefly on #archlinux-arm, and they said that they don't do automated testing
[19:45:18] <rewbycraft> Hm
[19:45:19] <tyzoid> but they have an automated build system
[19:45:21] <tyzoid> https://github.com
[19:45:25] <rewbycraft> I've seen that
[19:45:33] <rewbycraft> Maybe soemthing for deepthought to look at?
[19:45:39] <tyzoid> perhaps
[19:45:46] <tyzoid> though it's written in perl
[19:45:51] <tyzoid> and I've never touched perl before
[19:46:50] -!- WarheadsSE has joined #archlinux-ports
[19:46:59] <tyzoid> Hey WarheadsSE
[19:47:08] <tyzoid> rewbycraft: WarheadsSE is from the archlinux-arm project
[19:47:37] <rewbycraft> Ah
[19:47:38] <WarheadsSE> o/
[19:47:41] <rewbycraft> o/
[19:47:47] <WarheadsSE> xorond is also familiar with what we do
[19:47:58] <WarheadsSE> not to the grains that I am, but how the project functions
[19:48:25] <tyzoid> Yup. We had alerted him to the fact that arch's package maintainers are silently slowing down building i686 packages
[19:48:33] <tyzoid> effectively dropping support earlier than anticipated
[19:49:10] <tyzoid> but that's good to know
[19:49:25] <tyzoid> Did you get a chance to look through the thread I sent you?
[19:49:43] <WarheadsSE> not fully, but I have been somewhat aware of what was going on
[19:49:49] <tyzoid> okay
[19:50:04] <tyzoid> Basically the issue is that we had an iso fail to boot
[19:50:10] <tyzoid> and we're trying to avoid that from happening again
[19:50:21] <tyzoid> but also make sure that our limited staff can adequately handle making sure stuff works
[19:50:22] <WarheadsSE> Right
[19:51:10] <tyzoid> BTW, logs of this channe are kept on http://mirror.archlinux32.org
[19:51:15] <WarheadsSE> Well, we don't have testing built in, in the way you're looking for
[19:51:23] <WarheadsSE> Thanks for the heads up.
[19:52:03] <tyzoid> We had a couple of ideas, the main one was making a vm for doing package installation/functional tests
[19:52:10] <tyzoid> but writing those test cases can be a pain
[19:52:37] <WarheadsSE> That's true
[19:52:48] <tyzoid> Does archlinux-arm do manual testing by maintainers much in the way that the main archlinux project does?
[19:53:09] <WarheadsSE> We do, yes
[19:53:41] <WarheadsSE> but for most packages that are simply built from the AL trees without modification, we rely on them 'just working' and then fixing it when there is a build failure or a bug report.
[19:54:01] <tyzoid> WarheadsSE: We were thinking that too
[19:54:06] <tyzoid> but our user base is quite smaller
[19:54:12] <tyzoid> given that few people use x86_32
[19:54:44] <tyzoid> so it'd be more difficult to stay on top of broken versions of little-used packges
[19:54:53] <tyzoid> s/little/lesser/
[19:56:11] <tyzoid> rewbycraft: Did you have any questions about the build system?
[19:56:23] <tyzoid> I don't know any perl, so I'm not as useful there
[19:57:52] <rewbycraft> I'm a bit pre-occupied
[19:58:00] <rewbycraft> Am fixing someone's network
[19:58:09] <rewbycraft> And their network is fuckup badly.
[19:58:28] <tyzoid> alrighty then
[19:58:53] <tyzoid> WarheadsSE: The main person behind the build system is deep42thought, and he's away atm
[19:59:28] <tyzoid> If you're still around, I'll point him your way when he gets on
[20:01:33] <WarheadsSE> my weechat stays on almost 24/7
[20:06:28] -!- Polichronucci has quit [Ping timeout: 240 seconds]
[20:07:20] -!- Polichronucci has joined #archlinux-ports
[20:14:16] -!- Polichronucci has quit [Ping timeout: 260 seconds]
[20:15:19] -!- Polichronucci has joined #archlinux-ports
[20:19:35] -!- Polichronucci has quit [Ping timeout: 240 seconds]
[20:22:37] -!- Polichronucci has joined #archlinux-ports
[21:07:45] -!- Faalagorn has joined #archlinux-ports
[21:47:02] <brtln> tyzoid: which packages are not build for i686 and why none of you have told me? I'm here for a reason
[21:47:39] <brtln> you escalate to me, I shout at people, people in arch fix their stuff
[21:48:04] <tyzoid> brtln: I've only received vague statements
[21:48:10] <tyzoid> never gotten a list
[21:48:30] <tyzoid> If I find them, I'll let ya know.
[21:48:59] <tyzoid> and by vague, I mean generalized statements that implicate noone and everyone at the same time
[21:51:59] <rewbycraft> I can go poke on the buildserver and find a list
[21:52:15] <brtln> at some point, when I clone myself, we will have to talk about arch32 becoming tier2 arch, which at least would mean sharing some of mirrors and access to svn/git, don't want to be too specific because I haven't talked with anyone yet
[21:52:32] <rewbycraft> Is fair
[21:52:37] <rewbycraft> I'll see if I can fish up a list
[21:52:46] <rewbycraft> But I'm still not 100% on how deepthought's scripts work
[21:53:06] <brtln> well, I told you at the very beginning not to duplicate work and re-use alarm scripts :P
[21:53:21] <rewbycraft> Hey, deepthought's the brain behind the build system
[21:53:26] <brtln> WarheadsSE: also the same about ARM… I know Allan talked with you but I don't like the direction he took
[21:53:27] <rewbycraft> I just provide some server power
[21:53:41] <brtln> yeah, that was a general you
[21:55:04] <brtln> but at this rate of things to handle around Arch which everyone sees and no one notices, that's probably over my capacity of personal time
[21:55:24] <rewbycraft> Yeah, I get that
[21:56:10] <rewbycraft> I think this might be the list https://buildmaster.archlinux32.org
[21:56:56] * WarheadsSE reads up
[21:56:58] <brtln> maybe these are just broken? majority of the list is built for i686 too
[21:57:12] <rewbycraft> could be that the sources vanished or something
[21:57:33] <rewbycraft> The existance of a binary package in mainline doesn't mean we can actually build the package
[21:57:46] <brtln> never implied you can
[21:57:51] <rewbycraft> I know deepthought said there's a number where the PKGBUILD doesn't work anymore because the sources went missing
[21:58:13] <rewbycraft> Annoyingly, it doesn't put the logs for the broken packages on the buildmaster
[21:58:21] <rewbycraft> So I can't go and see *why* they're broken
[21:58:23] <brtln> yes, I was going to mention that
[21:58:28] <brtln> would be better to have logs
[21:58:31] <rewbycraft> Well, not without building them myself
[21:58:36] <brtln> still, I'm fine with you passing me a list of pkgbases
[21:59:11] <rewbycraft> Hey deepthought: Maybe make the build-slaves push the build logs to the master and shove them in a directory on nginx? Would probably be helpful when debugging
[21:59:53] <brtln> sweet mother of god, I love that haskell is now dynamically linked, but 400MB for one cli app is insane :(
[22:00:04] <rewbycraft> Yikes
[22:00:12] <rewbycraft> I mean, that's also why I dislike electron
[22:00:23] <brtln> becomes 690MB if I add git-annex to the list
[22:00:33] <brtln> at this point I can just use nixpkgs
[22:00:36] <rewbycraft> Because seriously... why do you need a full browser engine for every application
[22:00:41] <WarheadsSE> rewbycraft: our slaves push to the master, hence https://archlinuxarm.org
[22:00:54] <rewbycraft> Yeah, I've seen that before
[22:01:11] <WarheadsSE> gotta go start one some dinner. BBIAF
[22:01:12] <rewbycraft> As I said, I'm not the main person for the build stuff
[22:06:11] <rewbycraft> brtln: also, in case you haven't seen it, we've got a discussion about package testing going on the arch32 forums which you may have some insights on.
[22:12:39] <tyzoid> It's over on our forum at https://bbs.archlinux32.org
[22:13:45] <tyzoid> brtln: I'm not sure how tierings work for architectures
[22:13:48] <WarheadsSE> brtln: I don't understand your comment in regards to Allan
[22:14:31] <rewbycraft> tyzoid: Btw, I fixed the description on the system administration forum. It was bothering me. ;)
[22:14:51] <brtln> he had a dispute on merging efforts but maybe it wasn't you? anyway irrelevant at the moment
[22:15:27] <tyzoid> rewbycraft: Thanks
[22:16:55] <tyzoid> brtln: Afaik the broken packages are ones that fail their check() function
[22:17:14] <tyzoid> that is, of the ones we do have sources for
[22:17:27] <tyzoid> deepthought mentioned python as one of those
[22:24:11] <rewbycraft> It'd still be useful to have the actual build-logs
[22:30:49] <tyzoid> I don't disagree with that :)
[22:31:13] <rewbycraft> Yup
[22:32:32] <tyzoid> brtln: Do you have any comments/ideas for doing the automated testing in a sane way?
[22:38:11] <brtln> I will reply there tomorrow
[22:38:16] <tyzoid> sounds good
[22:38:45] <brtln> probably nothing close to rocket science
[22:38:57] <rewbycraft> Probably not
[22:38:58] <tyzoid> You're doing rocket science?
[22:39:24] * tyzoid wishes he was doing rocket science
[22:39:40] <rewbycraft> From what I've heard from the phsyics students here, it's hard as nails
[22:39:49] <tyzoid> Software isn't as bad
[22:39:54] <tyzoid> the hard science/math can be
[22:39:58] <rewbycraft> Oh yeah
[22:40:01] <rewbycraft> Definately
[22:40:03] <tyzoid> source: 3 years on the Michigan Mars Rover team
[22:40:08] <rewbycraft> Oh cool
[22:40:37] <rewbycraft> I'm still studying, but we had some fun with NASA plane GPS data the other day.
[22:40:41] <brtln> heh, I wish I did
[22:40:44] <tyzoid> The most difficult thing is doing teh control loops
[22:40:49] <tyzoid> s/teh/the
[22:41:07] <brtln> I mean, I doubt I will figure out anything you couldn't
[22:41:52] <tyzoid> I wouldn't discount your thoughts, though
[22:42:03] <tyzoid> Having more people thinking about a problem can't hurt
[22:42:03] <rewbycraft> An extra pair of eyes is always good
[22:42:06] <tyzoid> ^
[22:42:08] <rewbycraft> Especially when combined with a brain
[22:42:12] <tyzoid> ^ ++
[22:42:15] <rewbycraft> Because just eyes on their own are probably illegal
[22:42:19] <rewbycraft> Depending on your jurisdiction
[22:42:29] <tyzoid> um...
[22:42:34] <tyzoid> This took a turn
[22:42:54] <rewbycraft> Sorry, I've been in a very sarcastic discussion all day
[22:43:07] <tyzoid> no problem
[22:43:11] <tyzoid> I lol'd too
[22:43:25] <brtln> :D
[22:44:29] <rewbycraft> I was helping someone with a network that dropped off the internet... which wasn't fun
[22:44:38] <tyzoid> ah
[22:44:44] <tyzoid> That's resolved, I assume?
[22:44:46] <rewbycraft> And we managed to crash a server with "ip a show"
[22:44:55] <tyzoid> lol
[22:44:55] <rewbycraft> Not yet, waiting on a few people to get back to us
[22:45:06] <tyzoid> rewbycraft, do you remember my server load a few nights back?
[22:45:19] <rewbycraft> Wasn't it like 8 or something?
[22:45:39] <tyzoid> higher
[22:46:09] <tyzoid> Try several hundred
[22:46:11] <tyzoid> http://cdn3.tyzoid.com
[22:46:25] <rewbycraft> Woah
[22:46:29] <tyzoid> I was not a happy camper
[22:46:29] <rewbycraft> Nice
[22:46:36] <rewbycraft> I've seen 2900 when a disk died
[22:46:39] <rewbycraft> But daaannnggg
[22:47:05] <tyzoid> Minecraft decided it might be a good time to ignore my 'do not start if this file exists' file
[22:47:12] <tyzoid> so I had three minecraft servers running
[22:47:14] <rewbycraft> Wonderful
[22:47:18] <rewbycraft> Swapping?
[22:47:29] <tyzoid> Nah
[22:47:37] <rewbycraft> That's surprising
[22:47:38] <tyzoid> minecraft's disk usage is notorously bad, though
[22:47:51] <rewbycraft> I'm more worried about RAM usually
[22:47:57] <tyzoid> The system has several gigs of ram
[22:48:02] <rewbycraft> Fair
[22:48:03] <tyzoid> and I limited each to around 1-2g
[22:48:06] <rewbycraft> Ah
[22:48:07] <rewbycraft> Fair
[22:48:31] <rewbycraft> A few friends and I discovered a vps provider who shall not be named had a netdata instance open to the public
[22:48:51] <tyzoid> not familiar with netdata
[22:49:10] <rewbycraft> And one of their customer nodes were sitting with 8G in swap and 90+% RAM used and 22G or so deduped ram
[22:49:17] <rewbycraft> It's basically cpu, ram, disk monitorign
[22:49:41] <tyzoid> ah
[22:49:45] <tyzoid> kind of like cacti
[22:49:56] <rewbycraft> Yup
[22:50:09] <rewbycraft> But for real, the CPU, RAM and swap load was rediculous
[22:50:14] <rewbycraft> The definition of overselling
[22:50:43] <tyzoid> You might find this graph interesting
[22:50:45] <tyzoid> http://cdn3.tyzoid.com
[22:50:51] <tyzoid> It's a graph of available memory
[22:50:57] <tyzoid> not memory usage, as the title might suggest
[22:51:22] <rewbycraft> Not bad
[22:51:34] <tyzoid> Try and take a stab at what's causing the sawtooth graph of free memory
[22:51:40] <rewbycraft> I tend to assign <1-2G to my vms. But my hosts are all 16+G ram
[22:51:51] <rewbycraft> Also, rsync?
[22:51:55] <tyzoid> not rsync
[22:52:15] <tyzoid> this is a VPS I rent from a cloud provider
[22:52:19] -!- shentey has joined #archlinux-ports
[22:52:22] <rewbycraft> some cronjob, probably
[22:52:24] <tyzoid> nope
[22:52:30] <rewbycraft> Nothing on your server?
[22:52:34] <tyzoid> It is on my server
[22:52:39] <tyzoid> It's a daemon with a memory leak
[22:52:44] <rewbycraft> Wonderful
[22:52:48] <tyzoid> the spikes up are when oom-killer gets it
[22:52:54] <rewbycraft> Ah
[22:53:03] <tyzoid> but yea
[22:53:11] <tyzoid> it hasn't been affecting server performance too badly
[22:53:12] <rewbycraft> I was testing something and just allocated a new VPS on my setup with 8G of ram
[22:53:14] <rewbycraft> That felt good
[22:53:14] <tyzoid> so I haven't fixed it yet
[22:53:26] <tyzoid> that's nice
[22:53:39] <tyzoid> I've been wanting to get myself a dedicated server in a colocation facility for a while now
[22:53:48] <rewbycraft> I use OVH for my DC servers
[22:53:51] <rewbycraft> Mostly because they're cheap
[22:53:59] <rewbycraft> Although it's a slight hardware lottery
[22:54:01] <tyzoid> what's the ram on 'em
[22:54:07] <rewbycraft> Depends on what you go with
[22:54:18] <rewbycraft> You have their kimsufi brand for really cheap ones
[22:54:29] <rewbycraft> soyoustart for medium size (32G-64G ram)
[22:54:36] <rewbycraft> And their main OVH brand for 64G+
[22:54:52] <rewbycraft> Where the age of the hardware is in reverse order of the list I gave
[22:55:07] <rewbycraft> So kimsufi has the oldest hardware, soyoustart has the second oldest and OVH ahs the newest
[22:55:13] <tyzoid> figured
[22:55:26] <rewbycraft> (Basically, if the main OVH brand outdates it, SYS gets it, if SYS can't use it anymore, KS gets it)
[22:55:45] <tyzoid> That's a pretty good business model
[22:55:58] <rewbycraft> On the other hand, KS does 16G ram for 22 euro/mo
[22:56:03] <rewbycraft> So pricing's nice and cheap
[22:56:08] <rewbycraft> Although the hardware's a tad older
[22:56:24] <tyzoid> That's probably not an issue unless there's a lot of video encoding going on
[22:56:28] <rewbycraft> That's with 2T of storage and a i5-2300
[22:56:44] <rewbycraft> Yeah, I've got an E3-1245v3 in the server hosting my arch32 mirror and the buildmaster
[22:56:47] <rewbycraft> It does the job
[22:57:05] <tyzoid> Is that the server you run openstack on?
[22:57:08] <rewbycraft> Yup
[22:57:20] <rewbycraft> It's got 32G of ram, that cpu and 2x2T disks
[22:57:33] <rewbycraft> 250mbps guaranteed bandwidth with burst to ~0.8gbps
[22:57:41] <tyzoid> ooh nice
[22:57:41] <rewbycraft> And OVH's pretty good DDoS protection
[22:57:54] <tyzoid> my vps provider's ddos protection:
[22:58:02] <tyzoid> We've null-routed your server since there appears to be a ddos attack
[22:58:05] <rewbycraft> Heh
[22:58:11] <tyzoid> Me: >.<
[22:58:11] <rewbycraft> OVH runs it through their VAC system
[22:58:20] <rewbycraft> It takes a few hundred gbps before they start nulling
[22:58:35] <tyzoid> That's nice
[22:58:43] <rewbycraft> Yup
[22:59:10] <tyzoid> I was once helping someone out a few years back that was using OVH as a reverse proxy frontend for their ddos protectoin
[22:59:12] <tyzoid> protection*
[22:59:21] <rewbycraft> Yeah, it's pretty nice for that
[22:59:57] <rewbycraft> You're US right?
[23:00:00] <tyzoid> Yup
[23:00:04] <tyzoid> I'm over by Detroit
[23:00:09] <tyzoid> VPS is in chicago
[23:00:12] <rewbycraft> Ah nice
[23:00:23] <tyzoid> There's an internet backbone connecting here to chicago
[23:00:26] <tyzoid> so it's quite quick
[23:00:45] <rewbycraft> SoYouStart's got servers with 32G ram, 6c AMD and 2x3T disk for 39 usd/mo
[23:00:53] <tyzoid> That's not bad
[23:00:57] <rewbycraft> So but that's in their canada DC
[23:00:58] <tyzoid> and close to my budget
[23:01:02] <tyzoid> ehh
[23:01:11] <tyzoid> I'm close enough to canada for it not to matter too much
[23:01:17] <rewbycraft> They don't really have any major DCs outside of their canada and france DCs
[23:01:34] <rewbycraft> But yeah, they got some nice prices
[23:01:59] <rewbycraft> Although their customer service is a bit meeehhh
[23:02:05] <rewbycraft> Because it's all done through france
[23:02:11] <rewbycraft> And not all their reps speak english well
[23:02:14] <tyzoid> ah
[23:02:17] <tyzoid> that would stink
[23:02:18] <rewbycraft> So things often get lost in translation there
[23:02:30] <tyzoid> Do you usually interface with them in french?
[23:02:35] <rewbycraft> I rarely talk to them
[23:02:39] <rewbycraft> Mine's never really broken
[23:02:43] <rewbycraft> Once I had a disk fail
[23:02:49] <rewbycraft> Ticketed them with the SMART results
[23:03:01] <rewbycraft> Told them they're free to pull/replace it
[23:03:05] <rewbycraft> And they did the next day
[23:03:20] <tyzoid> How do you do backups of the server?
[23:03:29] <rewbycraft> Backblaze B2
[23:03:36] <WarheadsSE> Recently personally acquired 2x E5-2620 v1, new build box.
[23:03:41] <rewbycraft> Nice
[23:04:00] <rewbycraft> I like backblaze's pricing, 0.005usd/mo/gb
[23:04:02] <rewbycraft> Is not bad
[23:04:56] <tyzoid> It takes them 5 months to even make back the cheapest hdds
[23:05:09] <tyzoid> That's rediculously cheap
[23:05:22] <rewbycraft> They're fairly profitable as far as I've heard
[23:05:28] <tyzoid> I don't doubt that
[23:05:29] <rewbycraft> And they probably get major discounts
[23:05:34] <rewbycraft> Because they buy in massive bulk
[23:05:35] <tyzoid> since it's a subscription service
[23:05:38] <tyzoid> and that too
[23:05:56] <rewbycraft> the way they designed their system I think is to use consumer grade harddrives with software to handle the increased failure rate w/r to enterprise drives
[23:06:15] <rewbycraft> And then they just buy the drives by the pallet-load
[23:06:33] <rewbycraft> And they use custom server cases to cram as many drives in as little space as possible
[23:07:54] <rewbycraft> The design's open-source if you wanna take a look at it: https://www.backblaze.com
[23:08:03] <rewbycraft> 60 drives in a 4u chassis
[23:08:06] <rewbycraft> that's not bad at all
[23:08:27] <rewbycraft> And at the prices they charge, a singly machine's earned itself back in a year
[23:08:39] <rewbycraft> So yeah, not bad
[23:08:46] <tyzoid> That looks like a similar design to 45drives' 60-hdd enclosure
[23:08:56] <rewbycraft> Kinda, yeah
[23:09:02] <rewbycraft> Except they went a few steps further
[23:09:13] <rewbycraft> And don't leave the expansion room 45drives leaves for your server mobo and cards
[23:10:51] <rewbycraft> I love how they even put the price for the label on the BOM
[23:10:52] <tyzoid> at least the transition from scsi to sata has been good
[23:11:14] <rewbycraft> Yeah. I've got two scsi boxes sitting behind me
[23:11:30] <rewbycraft> They're my 32bit servers, and I still use 'em
[23:11:32] <tyzoid> scsi drives are quite expensive, now
[23:11:40] <rewbycraft> They take my low-cpu load stuff
[23:11:44] <rewbycraft> And are why I'm here
[23:11:55] <rewbycraft> (That and the collection of older gear I have in storage)
[23:12:08] <rewbycraft> Yeah, drives are getting expensive
[23:12:17] <rewbycraft> But I'm lucky these both have raid and should last a while
[23:12:22] <tyzoid> :)
[23:12:25] <rewbycraft> One of them's got 6 ultra320 scsi drives
[23:12:29] <rewbycraft> The other one's got 2 of 'em
[23:12:48] <rewbycraft> The one with 2 drives has two sun microsystems branded ones
[23:12:51] <rewbycraft> So damn they're old
[23:12:55] <tyzoid> My last encounter with scsi was a 1998 server in my high school
[23:13:08] <tyzoid> It was in a computer servicing class
[23:13:09] <rewbycraft> Ah nice
[23:13:18] <tyzoid> and the 2003 class had installed fedora onto it
[23:13:21] <tyzoid> that no longer worked
[23:13:51] <rewbycraft> Let me grab the 2-drive one and picture it for you
[23:13:57] <rewbycraft> I'm sure you'll enjoy a blast from the past
[23:14:05] <tyzoid> so I installed ubuntu-server, but found it didn't have support for ISA usb cards
[23:14:22] <tyzoid> that's be cool
[23:15:55] <tyzoid> I take that back, it was an ISA ethernet card
[23:16:07] <tyzoid> It was disappointing not to be able to get on the net with it
[23:16:30] <tyzoid> 728mb ram wasn't bad for a server from 1998
[23:17:53] <rewbycraft> https://dl2.pushbulletusercontent.com
[23:17:54] <rewbycraft> There ya go
[23:17:58] <rewbycraft> I put it on my bed
[23:18:16] <tyzoid> That's a 1u server
[23:18:19] <tyzoid> looks pretty good
[23:18:25] <rewbycraft> Yeah, made by sun microsystems
[23:18:32] <rewbycraft> That's a name from years past
[23:18:33] <tyzoid> I like the VGA port in the front
[23:18:35] <rewbycraft> It's got a Xeon in it though
[23:18:38] <tyzoid> I remember sun ms
[23:18:45] <rewbycraft> Most my rack gear has front-VGA
[23:18:56] <rewbycraft> That's a pretty common feature
[23:19:08] <rewbycraft> It's so you can come by with a monitor and keyboard on a trolley and just plug in
[23:19:17] <tyzoid> I remember doing development for a solarus spark machine, still running a sun-ms branded os
[23:19:27] <tyzoid> that was back in 2015
[23:19:30] <rewbycraft> Oh dang
[23:19:32] <rewbycraft> solaris, probably
[23:19:35] <tyzoid> yup
[23:19:43] <rewbycraft> Cool
[23:19:49] <rewbycraft> This thing can probably run that
[23:19:55] <tyzoid> libc was 15 years old on that machine
[23:20:03] <rewbycraft> But I have arch on it at the moment
[23:20:16] <rewbycraft> Surprisingly, kernel 4.9 still happily boots it
[23:20:27] <rewbycraft> And it's got 4G of ram and dual dualcore xeons
[23:20:34] <rewbycraft> And it's pretty quoet
[23:20:36] <tyzoid> probably because xeon had some advanced features that didn't make it into the consumer processors for years
[23:20:37] <rewbycraft> *quiet
[23:20:40] <rewbycraft> So it's still useful
[23:21:00] <rewbycraft> I was more worried about the NICs
[23:21:04] <rewbycraft> But they work just fine
[23:21:14] <rewbycraft> The SCSI HBA's fine too
[23:21:35] <rewbycraft> I use it for test work
[23:21:42] <rewbycraft> May make it into a build-server for arch33
[23:21:44] <rewbycraft> *arch32
[23:22:01] <rewbycraft> So we have some packages built on actual 32bit machines
[23:22:15] <tyzoid> can't openstack fire up 32bit vms?
[23:22:20] <rewbycraft> It can
[23:22:23] <rewbycraft> But what's the fun in that
[23:22:34] <rewbycraft> It just launches KVM vms in my setup
[23:22:37] <tyzoid> I think doing the testing on actual hardware is more important than the building
[23:22:38] <rewbycraft> Can easily install 32bit arch on that
[23:22:42] <rewbycraft> That's possible too
[23:22:46] <rewbycraft> I can donate it to testing
[23:22:47] -!- shentey has quit [Quit: Konversation terminated!]
[23:23:01] <rewbycraft> I'd only run this a few hours a day though. Power-usage is mad on it
[23:23:05] <rewbycraft> 300W or something
[23:23:22] <tyzoid> That's more than a recent gaming build of mine
[23:23:30] <rewbycraft> I could probably make it PXE boot by default
[23:23:44] <rewbycraft> And then we can just run whatever tests we need from some network drives
[23:23:49] <rewbycraft> (because the scsi drives are slow)
[23:24:02] <tyzoid> hey rewbycraft, do you do any C programming?
[23:24:03] <rewbycraft> And then it can just sit turned off if it's not doing any testing
[23:24:06] <rewbycraft> I do, yeah
[23:24:16] <tyzoid> I've got a project that I've been trying to get started on
[23:24:19] <rewbycraft> (It does WoL, I checked)
[23:24:23] <tyzoid> and I've got some time to work on it now
[23:24:34] <tyzoid> It's a new implemtation for a networked filesystem
[23:24:38] <rewbycraft> I'm revising some stuff that needs to be handed in before midnight
[23:24:47] <rewbycraft> So if you wanna get into a large discussion, please wait 30 minutes
[23:24:51] <tyzoid> sure thing
[23:24:56] <tyzoid> ping me when you're done
[23:25:05] <rewbycraft> Will do
[23:35:01] <rewbycraft> tyzoid: ping
[23:35:05] <rewbycraft> took a lot quicker
[23:35:10] <rewbycraft> I was expecting to have to chase someone down
[23:35:11] <tyzoid> nice
[23:35:12] <rewbycraft> Didn't have to
[23:35:45] <tyzoid> So as I was saying, my project is a new way to have distributed file systems
[23:35:47] <rewbycraft> Maybe we should take in a PM, not sure if deepthought wants such a long (and unrelated to arch32) discussion here
[23:35:51] <tyzoid> sure
[23:35:51] <rewbycraft> *talk
[23:36:01] <rewbycraft> Feel free to pm me
[23:36:06] <tyzoid> not sure how
[23:36:11] <tyzoid> I'm not familiar enough in weechat
[23:36:16] <rewbycraft> Do /query rewbycraft
[23:36:21] <rewbycraft> And that opens a buffer for the pm
[23:36:26] <rewbycraft> And you can just type there like a normal chat
[23:36:30] <tyzoid> thanks
[23:43:46] <WarheadsSE> Ohai, it's r00t^2 and drathir too
[23:43:57] <tyzoid> lol
[23:44:09] <r00t^2> indeed
[23:44:19] * WarheadsSE pokes xorond
[23:44:40] <r00t^2> WarheadsSE: that reminds me, did you see that revival of AIF i wrote? acts more like kickstart than the original AIF
[23:44:44] <WarheadsSE> Just checked the user list while I'm a passenger
[23:45:02] <r00t^2> kickstarter?
[23:45:05] <r00t^2> whatever rhel calls it
[23:45:10] <WarheadsSE> See/hear/watch jthan get pissed?
[23:45:13] <r00t^2> not the crowdfunding thing
[23:45:21] <r00t^2> he's always pissed
[23:45:29] <WarheadsSE> Yup
[23:45:38] <WarheadsSE> Kickstart*
[23:45:56] <r00t^2> thanks. i know how to write the configs for them but can never remember the name of the tech
[23:45:58] <WarheadsSE> I have heard but haven't had the full time to review yet
[23:46:09] <r00t^2> https://aif.square-r00t.net
[23:46:20] <r00t^2> there's a config generator too, it's not documented there but it is at... sec
[23:46:42] <r00t^2> https://wiki.archlinux.org and https://wiki.archlinux.org
[23:47:00] <WarheadsSE> Yup, just not time.