IRC Logs for #crux-devel Monday, 2012-05-21

teK_frinnst: yeah kqemu should be DISMISSED01:31
frinnstanybody feels like going over the orphaned ports and nuke most of them?01:33
frinnstlooking at the list last night made me mad :)01:33
teK_because it's long?01:34
frinnstyeah and all the old shit that doesnt even need to be in opt01:34
teK_IF we nuke them, there should be a post to the ML imho01:35
frinnstwell sure01:35
teK_it's one of the small things that are suboptimal with CRUX, we do not have a 'beautiful' mechanism of notifying users about these changes01:35
teK_like gentoo's message system01:35
frinnstyes. unless they are subscribing to the ml they are kind of screwed01:36
frinnstbut honestly, I doubt most ppl will even raise an eyebrow regarding these ports01:36
frinnstone or two exceptions perhaps01:37
frinnsti guess quite a few ports depend on libmad. guess that should stay01:37
teK_I remember dl'ing cedega (the 'full' version) from some dark-unofficial mirror01:38
teK_oh noes, facebook shares may be have been overrated. no shit sherlock.01:39
tilmanhopefully that shit tanks01:45
teK_explain that to me..01:47
teK_why for heaven's sake is linux 3.4 mainline and not 'stable'01:47
teK_it's impossible to get the newest official version out of
teK_and it's only available as a patch01:48
Romsteri've been doing a orphan sweep of opt and contrib after seeing what dependencies need to stay02:52
Romsteri also have asked prologic a number of times if he'll pick up myht tv stuff in contrib.02:53
frinnstpleeease let me be the one that purges emacs from opt (vi vi vi - the editor of the beast ftw!)04:09
* frinnst just spent 2hrs at a fish-plant04:09
frinnsti smell like shit04:10
frinnstrather, like fish04:10
* prologic wants to maintain mythtv in contrib04:11
* prologic but lacks motivation to learn git well04:11
frinnstgit add mythtv && git commit -m "mythtv: new maintainer" && git push04:11
frinnstyou dont need to know much04:12
*** horrorStruck has joined #crux-devel05:09
*** joe9 has quit IRC05:53
*** joe9 has joined #crux-devel05:55
*** mike_k has joined #crux-devel07:17
Romsterprologic, git fetch ; git rebase origin/2.7 ; do your edits... ; git add files... ; git commit -m "some message" ; git push07:19
Romsterand if someone beats you to a commit before you get to git push just do "git fetch ; git rebase origin/2.7" again before git push07:19
Romsteronce the git tree is cloned and configured
jaegerteK_: what do you need? was asleep07:34
teK_I had a queston about vmware, got a minute?07:35
teK_is the vCenter Server and vSpehre thingy just a namechange or what's the difference between them?07:37
jaegervSphere is the overarching name for the entire virtualization suite07:38
jaegerso vcenter, ESXi, etc. are part of vSphere07:38
jaegervcenter is the management server part of it, it consolidates your ESXi hosts into one (or more) manageable datacenters07:38
jaegerif you don't use vcenter you have to connect your vsphere client to each ESXi host individually07:38
jaegerif you only have one host that's no problem07:38
jaegervcenter is also required for enterprise features like FT, HA, vMotion, Storage vMotion07:39
teK_an scheduling of vmware converter tasks, right?07:39
teK_because this friggling converter does has task capability but not in standalone (read: free) mode07:40
jaegerI haven't used the converter often but that wouldn't surprise me07:41
teK_I test-installed (interprete as you like) vCenter and the only thing I got to see was the orchestrator (wtf) and a list to enter 'licensed/managed' ESXi hosts07:43
teK_all I want is this scheduling thingy of the converter, no datacenter management no vMotion no nothing; just conversion on a regular basis07:43
teK_but I guess you can't tell me what to do in that case07:43
jaegerPerhaps you can script it with the CLI tools07:50
jaegerI've only used converter once that I can recall, so I can't really offer much in the way of help with it, unfortunately07:50
teK_nope, afaics07:59
teK_additionally it b0rks hard while cloning C:+E: partitions (after C:..): mounted partition cannot be formatted (????)08:00
teK_of course not ;)08:00
teK_wrt CLI08:00
teK_thanks for your help then08:04
jaegerWish it were more useful08:09
teK_well I found my love for qemu again but the problem that remains to be solved is cloning08:09
teK_but there's other commercial products08:09
jaegerNo way to clone with qemu?08:10
teK_a physical installation of windows?08:11
teK_how should that work out ;)08:11
jaegerWindows is required to clone?08:11
jaegerOr do you mean you need to convert a physical windows installation to a qemu image?08:12
jaegerAny reason you can't use qemu-img to convert the physical disk to a qcow/qcow2 file?08:15
teK_what about drivers? IDE vs S-ATA vs SCSI etc?08:17
teK_I want to clone 3-5 machines on a regular basis to run a shadow data 'center' of our production environment for emergency cases08:17
jaegerWhat I would try is this:08:19
jaegerrun sysprep on the original installation to remove drivers, etc.08:19
jaegerclone it to a 'master' qemu image08:19
jaegeruse that 'master' image to create all the others later08:19
teK_but that won't work with a running windows, I'd have to restart the servers with eg clonezilla etc08:20
jaegeryes, but only once08:20
teK_and they're unique machines so no master image ;)08:20
jaegerwell, the 'master' image is just a sysprepped image08:21
teK_how should that work?08:21
jaegerwhen you boot it windows will attempt to install all the "new" hardware it finds08:21
teK_i.e. mergin the newer version with the qemu-master08:21
teK_yeah I read of sysprep08:21
teK_it's what vmware does, too08:21
jaegerwell, how often do you need to do this? If it's not too much work I'd just manually make changes to both... unless you can schedule downtime for a new clone regularly08:21
jaegeror replace your physical stuff with virtual stuff permanently :)08:22
teK_daily would be best08:22
teK_changes to both? copy files? I don't expect that to work08:22
teK_yeah that would be step 208:22
jaegerI obviously don't understand what you're asking08:22
jaegerWhat do you mean by "i.e. mergin the newer version with the qemu-master" ?08:23
teK_you said: create a master image then 'create all others later'08:23
jaegerok, let's say you have 5 unique physical machines08:24
teK_how would the 2nd step work if there where changes made to system parts (bootloaderconfig/updates/ ...)08:24
jaegeryou take them all down and make 5 qemu-img images08:24
jaegerthen when you need clones in the future, you clone those images08:24
jaegernot the original physical hardware08:24
teK_I need current copies of the physical machines08:25
jaegerIf your entire purpose is to back up the physical hardware daily then you probably should use backup software instead of vm cloning08:25
jaegerOr pay for a commercial product that would allow you to do it without downtime08:25
jaegerif there is such a thing08:25
teK_our desaster scenario is: all physical (original) servers are gone08:25
teK_vmware converter works partially and without downtim e;)08:26
jaegerOk, so back to figuring out how to script it :)08:26
teK_it uses a sqlitedb for its tasks08:26
jaeger <-- perhaps this is helpful?08:26
teK_I rev engeinered "" it with triggers =)08:27
teK_I know that site, 1st hit on google :P08:27
teK_newer version of the converter won't let you generate/import XML files, IIRC08:28
teK_but I will recheck08:28
teK_thanks for the hint08:28
teK_I tried a while ago so I don't quite remember08:29
jaegerfor what it's worth I figured out the source of my confusion earlier... when you said cloning I assumed you meant you wanted copies of servers that would also be running, hence clones08:42
jaegernot backups :)08:42
teK_you're right backup would be the more appropriate term08:45
juefrinnst: thanks09:25
teK_hi jue09:25
teK_Romster: thanks for the prompt patching, will test tonight :}09:26
juefrinnst: feel free to remove the old cruft, IMO ;)09:27
jaegerheyo, jue09:31
frinnstlibmad is the only port i see that should remain in that orphaned list09:36
frinnstpossibly openldap if someone uses that09:36
frinnstunsure if anything would break if we also purged those p5-* ports listed09:36
* teK_ thinks openldap shall stay09:38
Romsteri probably should take libmad since i use that in other ports i maintain then again i've been picking up far too many ports lately. would be nice if others helped out to even the number of ports maintained per a maintainer.09:43
frinnstI can pick up libmad, no worries09:43
RomsterteK_, i haven't noticed anything strange since patching and i did have that crazy bleed though before patching i thought my monitor was having a spaz attack09:43
Romsteras i had text over a flash video and it made the black text appear white in one area of the screen only on the edges though like aliasing gone horribly wrong.09:44
teK_no, it's gone if I switch to html5 videos09:44
Romsterof course html5 doesn't use flash09:45
*** tilman has quit IRC09:45
*** tilman has joined #crux-devel09:47
*** horrorSt1uck has joined #crux-devel09:49
Romsterhmm i seem to be using openntpd09:49
teK_user ntpd :}09:52
*** horrorStruck has quit IRC09:52
teK_or ntpdate09:52
Romsteri've been putting it off but i should pick up cabextract openntpd xorg-font-msttcorefonts09:52
Romsteri've had openntpd setup and working from years ago it works why change now?09:53
Romstereffort xD09:53
teK_jaeger: if I don't get this wrong the p2vtool from the tutorial is not included in recent vmware converter versions :-)09:53
Romstermaybe if i was redoing everything from scratch09:53
teK_because you'd not have to maintain the port.09:54
jaegerteK_: doh09:54
teK_well, $$$ :)09:54
Romsteri don't think it's ever been updated so no effort other than picking it up, but i gues i should get unlazy and break my existing ntpd setup to use something more maintained.09:55
Romsterhmm 2009 newer version of openntpd too... ok it's not as old as i was expecting...09:57
jaegerIt very rarely needs to change, not like NTP is really evolving quickly09:58
Romsteryeah i'd expect it to be like ~3-4 years between any updates.09:58
Romsterdebating if ntp in contrib is a better choice or keeping to the one i'm on now.09:59
Romsterdecide later, i think i got this coming weekend to myself for a change than working every weekend.10:02
Romsterfrinnst, there /was/ a ton more unmaintained before i attacked that list.10:03
* jue points at chrony ;)10:09
jaegergrrr... any udev expert around? on my latest updated ISO device nodes aren't being created for partitions10:14
jaegerI create a partition layout for sda in fdisk and then sdaX devices don't get created10:14
Romsterjaeger, you tried have udevadm trigger and udevadm settle?10:16
jaegeryeah, didn't help10:16
Romsterno errors in dmesg about udev rules?10:17
Romsterthere should be some rule to make disk nodes... when the kernel sees new hardware... hotplug ahci or was it acpi too many terms in my head.10:18
Romsteri guess you already reloaded the partition table to kernel.10:19
jaegeryeah, it's correct. just no device nodes until a reboot10:19
Romsterah yeah i've came across that before, something stupid about the kernel wont reload the partition table as it's in use hits some ioctl error.10:20
Romsteri've seen many users in #lvm hit this same issue too.10:20
jaegerthat's not the problem here, the kernel sees the proper partition table10:20
jaegerudev just isn't creating the new devices10:21
Romsteri don't know the fix for that other than a reboot since some process deadlocks in the kernel and nothing fixes it.10:21
Romster/lib/udev/rules.d/ diff -u on that to another machine that is working ok?10:22
Romster/lib/udev/rules.d/60-persistent-storage.rules should be the file responsible for that10:23
jaegerThere aren't any custom rules, it's installed from the default udev package10:24
Romsterunless it's been hand edited or messed up by keep during a rejmerge.10:24
jaegerthere are no rejmerge runs10:24
jaegerIt's the updated ISO, it's bone stock10:25
Romsterif that's even possible, hmm no it's not tracked only the /etc is...10:25
Romsterhmp well i'm lost. other than probably a bug or kernel config.10:25
Romsterdo you see anything with blkid even10:25
Romsterif that lists the partiton you created then i'd be thinking that udev didn't see the event the kernel was meant to tell it new disk to rescan or something.10:26
jaegeronly the cdrom device, and ata_id exits with a 1 status, though it prints nothing at all.. very odd10:27
Romsteri'm not udev expert but i had poked around in that area in the past.10:27
Romsterenable debugging in kernel/udev10:27
Romstertry older udev might be some bug?10:28
Romsterbtw i know this isn't the case but if oyu happen to know any way to get around that deadlock ioctl kernel issue on re-reading partition table, please let me know how. it's one very irritating issue i constantly hit and other users in lvm.10:34
jaegerI doubt there's any way10:35
Romsternot fun telling them i know of no other way than a reboot and some of them run production servers, can live migrate data in lvm but if you have a PV in a real partiton and you grow the partiton then try to re-read it so you can use pvresize. you can't even re-read the larger partition table.10:35
jaegerrescanning the scsi bus probably won't help for lvm devices10:36
*** horrorSt1uck has quit IRC10:36
Romsternot when the kernel errors on some ioctl busy...10:36
Romsteri tend to use whole disks so i rearly have that issue myself.10:37
juejaeger: sorry, no really a good idea about the issue10:42
juejaeger: do you have a /run directory?10:43
jaegertrying to dig through it now10:43
Romsterlet us know what you find so we all get more wise10:43
jaegerI doubt it, haven't created it manually... thought that udev would automatically fall back if it didn't find /run10:43
Romsteri better get to bed g'night10:44
Romsteroh i saw a recent ML entry about do not create /run didn't read why though.10:44
juehmm, not sure about that. IIRC udev > 175 needs a /run dir to work correctly10:44
juenight Romster10:44
Romsterlook in crux-lists ttp://
jaegeractually it probably is created, it's part of the udev port now, right?10:45
jueyeah, should10:45
jaegeryeah, it's there, I just booted it again to check10:45
Romsternow i'm off :)10:46
frinnstoh right. I dont have libmad installed / no use for it12:22
frinnstguess it can be hard to maintain then12:22
frinnsthmm, seems a few ports on the orphaned list has already been picked up (p5-class-std, openntpd, p5-berkeleydb, p5-io-multiplex)12:41
*** mike_k has quit IRC15:24
*** hawksbill has quit IRC15:41
*** hawksbill has joined #crux-devel15:42
*** pitillo has quit IRC17:46
*** pitillo has joined #crux-devel17:48
*** joe9 has quit IRC19:38
*** joe9 has joined #crux-devel19:52
*** __mavrick61 has quit IRC21:36
*** ___mavrick61 has joined #crux-devel21:37

Generated by 2.11.0 by Marius Gedminas - find it at!