Monday, 2017-01-09

*** toan has quit IRC01:23
*** zerick has quit IRC01:23
*** zerick has joined #openstack-powervm01:23
*** toan has joined #openstack-powervm01:25
*** thorst has joined #openstack-powervm01:39
*** thorst has quit IRC01:44
*** thorst has joined #openstack-powervm01:56
*** thorst has quit IRC02:09
*** toan has quit IRC02:43
*** toan has joined #openstack-powervm02:47
*** edmondsw has joined #openstack-powervm02:53
*** edmondsw has quit IRC02:57
*** apearson has joined #openstack-powervm03:25
*** thorst has joined #openstack-powervm05:38
*** thorst has quit IRC05:43
*** apearson has quit IRC05:56
*** k0da has joined #openstack-powervm06:53
*** thorst has joined #openstack-powervm07:39
*** thorst has quit IRC07:43
*** thorst has joined #openstack-powervm08:00
*** thorst has quit IRC08:04
*** openstackgerrit has quit IRC08:18
*** jwcroppe has quit IRC08:48
*** AlexeyAbashkin has joined #openstack-powervm09:44
*** thorst has joined #openstack-powervm10:00
*** k0da has quit IRC10:02
*** thorst has quit IRC10:07
*** k0da has joined #openstack-powervm11:15
*** smatzek has joined #openstack-powervm12:06
*** thorst has joined #openstack-powervm12:24
*** thorst has quit IRC12:24
*** thorst has joined #openstack-powervm12:25
*** kylek3h has quit IRC13:07
*** edmondsw has joined #openstack-powervm13:25
*** mdrabe_ has joined #openstack-powervm13:27
*** kylek3h has joined #openstack-powervm13:34
*** thorst has quit IRC13:45
*** thorst has joined #openstack-powervm13:46
*** viclarson has joined #openstack-powervm13:50
viclarsonhi dudes13:50
*** thorst has quit IRC13:50
*** svenkat has joined #openstack-powervm13:51
*** apearson has joined #openstack-powervm14:00
*** efried has joined #openstack-powervm14:01
*** thorst has joined #openstack-powervm14:05
thorstefried adreznec: looks like the CI has held up OK14:10
*** tblakes has joined #openstack-powervm14:14
*** dwayne__ has joined #openstack-powervm14:18
*** tlian has joined #openstack-powervm14:18
*** mdrabe_ has quit IRC14:27
*** mdrabe_ has joined #openstack-powervm14:27
*** smatzek has quit IRC14:30
viclarsonafter start vm port response as vnc and port external_ip:5900 response as http14:35
*** efried has quit IRC14:36
viclarsondoes external_ip:5900 emulate novnc-proxy and14:36
viclarsoncan be linuxvnc service configured to listen external_ip?14:37
*** jwcroppe has joined #openstack-powervm14:46
*** efried has joined #openstack-powervm14:51
*** kriskend_ has joined #openstack-powervm15:00
*** smatzek has joined #openstack-powervm15:01
thorstviclarson: sorry...just saw this15:12
thorstthis is using the OpenStack nova noVNC bits, or just straight pypowervm?15:12
efriedthorst: quick discussion with svenkat about SR-IOV migration?15:20
*** mdrabe_ is now known as mdrabe15:23
efriedthorst: We're looking at the compute-side plug method.  On a migration, the bridge mappings (pport loc codes) are coming from the source system - so they're no good on the dest system.15:24
efriedSo at least in the migration case, we have to go fetch the msys and get new pport-to-label mappings.15:25 the plug method.15:25
thorstsvenkat mentioned that to me on Friday15:26
svenkatand also, I am not able to distinguish between deploy / migration with respect to plug . so at this point, retrieval of port mappings needs to be done on compute side for all scenarios (deploy and migration)15:26
thorstsaid he was going to call neutron to do that.15:26
efriedIt would be neat if, in the non-migration case, we could do the (cheap) dict lookup from the bridge mapping we got from the neutron side; and only do the (more expensive) REST call to rebuild those mappings if we're in a migration situation.15:26
svenkatthorst : yes. that is for retrieving physical network name for a given network.15:27
efriedWe're not even sure it's possible for the plug method to tell we're migrating.15:27
efriedAnd realistically, even if we could, the set_vnic_backdevs method has to look up the msys anyway (unless the wrapper is passed in).15:27
efriedSo the proposal is this:15:28
efriedGet rid of the bridge mapping entirely on the neutron side.  Always grab the msys on the plug side; and pass it into set_vnic_backdevs.15:28
thorstwhat does the agent do now?15:28
efriedAdded bonus, we reduce/eliminate the possibility of the timing window where somebody changed pport labels between when the neutron side built the bridge mapping and when plug is called.15:28
efriedThe agent does the msys lookup, builds the pport/label mappings, and stuffs 'em into (something that eventually comes through to the plug method as) the 'vif' object.15:29
*** tlian has quit IRC15:29
efriedHonestly, I can't remember why we thought that was a better idea than doing it on the plug side.15:29
thorstI think it was for multiple vifs or something15:30
thorstlookup once...15:30
thorstI dunno, anyway, I'm OK with it.  Is svenkat driving that change?15:30
*** tlian has joined #openstack-powervm15:30
svenkatyes..but i have another point, lets leave the bridge mapping being generated and maintained in vif . the reason being a potential future work that involves direct vf for plug15:31
thorstwhy would leaving it there matter?  Why not just look up and keep one common code path?15:31
svenkatdont you think we will get benefited by getting port mapping from agent for direct vf - it will be advantageous interms of performance15:31
thorstits a 1 second call.15:32
svenkatsriov agent and its mechanism driver will do the work and compute need not have to lookup - for direct vf where migration is not a concern.15:32
thorstI just ran it.15:32
svenkatyes. it is quick.15:32
thorstI'd rather get rid of the debt15:32
svenkatok, i will take it off.15:32
thorstgiven that this is a rare scenario15:32
thorstif this was hot path...different story15:32
efriedYeah, that gets rid of 80% of the agent code.15:32
svenkatI agree.15:32
svenkati will get these done and send out a patchset for both nova and networking_powervm15:33
efriedOkay, thanks thorst and svenkat15:33
svenkatefried , thorst, how about this is netwoking_powervm change, can it be merged? i see a +2 from efried15:35
thorstJust remove the bug?15:36
thorstits not a bug?15:36
thorsto wait, you need to merge it back to newton...15:37
thorstI just really don't like the bug report on that thing...but I'm OK merging it15:37
thorstI guess I defer to efried...if he's OK with it, then we can just W+115:37
efriedisn't that code going to go away now?15:38
efriedoh, I guess not.15:38
svenkatefried.. no it is is needed.15:38
efriedMerge away.15:38
*** viclarson has quit IRC16:04
*** AlexeyAbashkin has quit IRC16:16
*** efried has quit IRC16:31
*** efried has joined #openstack-powervm16:43
*** k0da has quit IRC16:56
*** edmondsw has quit IRC17:04
*** edmondsw has joined #openstack-powervm17:05
*** efried has quit IRC17:06
adreznecthorst: I've theoretically just finished the dns mirror and apt cache playbook code (sets it up and makes changes to all potential clients to use the mirror/cache)17:43
adreznecGoing to test it after lunch here :P17:43
adreznecIf it works we'll need to rebuild the dsvm nodes17:43
thorstadreznec: do we want to wait on the rebuild until esberglu is back?17:52
*** jpasqualetto has joined #openstack-powervm17:57
*** thorst has quit IRC18:30
*** thorst has joined #openstack-powervm18:31
*** thorst has quit IRC18:35
*** thorst has joined #openstack-powervm18:49
*** k0da has joined #openstack-powervm19:17
*** edmondsw_ has joined #openstack-powervm19:20
*** edmondsw has quit IRC19:22
adreznecthorst: I'll probably wait on the client side changes19:32
adreznecDoing the server tests though since those are pretty standalone19:32
adreznecI've theoretically written the code for it all though19:32
adreznecAssuming it works :)19:32
thorstI'm just thinking we can get esberglu to get heads down on it when he gets back19:41
thorstcause he's got a bunch of patches that were part of last deploy19:41
thorstI worry about brittleness19:41
adreznecUh huh19:41
adreznecThere are a lot of things deployed but not merged19:41
adreznecNoticed that while messing with zuul last week19:41
thorstwould be impossible for qing wu to do any redeploy19:42
adreznecWe could probably figure it out, but it would be... painful19:42
thorstand with things stable...we can wait a week19:45
thorstthe key is you got everything primed...and in a week or two it should be way faster.19:45
adreznecNote to self: attempting to use the "openstack-ansible" command to run the CI system ansible stuff doesn't actually work19:49
*** efried has joined #openstack-powervm20:05
*** jpasqualetto has quit IRC21:13
BorD__anyone get yum to work on aix?21:17
BorD__says first step is to update rpm, but there are a ton of failed dependencies21:17
*** svenkat has quit IRC21:22
*** smatzek has quit IRC21:23
adreznecBorD__: I assume you're following something like ?21:27
adreznecHmm ok. I haven't done it but I know one of the guys on my team has. Let me ping him quick21:28
BorD__thanks, just seems like if you have a bunch of dependencies then there should be an ez bundle for whatever you need for rpm too21:29
BorD__ie here we need to update rpm but we are in rpm hell to update rpm in order to get yum to work21:29
adreznecYep, asking around here21:33
BorD__lord.  complains it wants gettext <= 0.17, and gettest-0.19 is what is on the toolbox site.21:47
BorD__gettext-0.19 even21:53
adreznecBorD__: What AIX version are you trying to do this on? We'd last done it on AIX 7.1 TL4 without issues21:55
*** burgerk has joined #openstack-powervm21:55
BorD__new install of aix21:56
BorD__just downloaded the rpm from the toolbox, doing an rpm -ivh21:57
BorD__and get a bunch of failed dependencies for rpm21:57
BorD__rpm -
burgerkis it the rpm.rte that you are having issues with ?21:59
BorD__just trying to update it so i can install the yum bundle22:00
burgerkHere is what I did ...  download rpm.rte (not one of the specific versions)22:00
BorD__not one of the specific?  what do you mean?22:01
burgerkthe instructions point to
burgerkthere are serveral versions there along with one that is unversioned ( in the name )22:02
BorD__aaargh..  gotcha...22:02
burgerkrpm.rte                                22-Jul-2016 05:02   44M22:02
burgerkcd to the directory where that rpm resides and do the following:22:03
burgerkrm .toc22:03
burgerkinutoc .22:03
burgerkinstallp -acgXYd . rpm.rte22:03
BorD__that must be my whole problem.  i can freaing read22:03
BorD__thank you so much for reading for me.  let me try that file...22:04
BorD__been too long of a day22:04
BorD__and you did that instead of the
burgerkyes that is what worked for me :)22:05
BorD__thanks, downloading now.  slow link.22:05
BorD__appreciate the offtopic help22:06
burgerklet me know if you still see issues22:19
BorD__still downloading22:20
BorD__slow link.  close though22:20
BorD__although what is the point of having rpm in both the rpm section of the toolbox and the installp section if the installp section is what you want?22:21
BorD__appears to be working..22:26
BorD__the installp is chugging along..22:26
adreznecCool, hopefully things work as designed22:28
BorD__ha.  now the gettext issue.22:31
BorD__gettext-0.19 is installed but it wants gettext-0.1722:31
BorD__what is AIX-rpm-
*** tblakes has quit IRC22:33
burgerkis this part of the rpms from the yum_bundle or still the rpm.rte ?22:34
BorD__rpms in general.  i wanted to remove gettext and let it install from the bundle22:35
BorD__but i had not seen this AIX-rpm22:35
BorD__not sure if it just came in with the new rpm?22:36
burgerknot sure22:36
burgerkI have done installs of yum but am not an expert on all the AIX details behind it22:37
BorD__im just trying to uninstall rpms22:37
BorD__but everything says it is needed by this AIX-rpm22:37
burgerkif I remember right, I tried installing the yum rpm first and then just worked my way backward on dependencies vs. just installing everything in the bundle22:38
BorD__problem is, i have gettext.0-19 and i cannot uninstall it22:39
BorD__so i cant install the gettext from the bundle22:39
burgerkwhich is requiring gettext?  yum itself ?22:40
BorD__a specifi version that is older than this one22:40
BorD__i need to just nuke rpm and start over.22:43
burgerkI would get all the other dependencies and when it is just that one left, ignore it22:43
BorD__holy cow i think that did it22:46
*** kylek3h has quit IRC22:51
*** burgerk_ has joined #openstack-powervm22:52
*** edmondsw_ has quit IRC22:55
*** burgerk has quit IRC22:55
BorD__man i can win.  yum install samba-client, works like a charm but dont see smbmount22:55
*** burgerk_ has quit IRC22:57
*** dwayne__ has quit IRC23:00
*** apearson has quit IRC23:15
*** kriskend_ has quit IRC23:24
*** openstackgerrit has joined #openstack-powervm23:32
openstackgerritDrew Thorstensen (thorst) proposed openstack/nova-powervm: Change namespace to nova.virt.powervm_ext
*** thorst has quit IRC23:32
*** thorst has joined #openstack-powervm23:32
*** thorst has quit IRC23:37
*** efried has quit IRC23:46
*** k0da has quit IRC23:51
*** mdrabe has quit IRC23:51
*** edmondsw has joined #openstack-powervm23:57

Generated by 2.14.0 by Marius Gedminas - find it at!