openstackgerrit | Merged openstack/ironic-python-agent master: Use lshw in place of dmidecode for the default hardware manager https://review.openstack.org/501799 | 00:00 |
---|---|---|
*** hshiina has joined #openstack-ironic | 00:03 | |
*** rcernin has joined #openstack-ironic | 00:08 | |
openstackgerrit | James E. Blair proposed openstack/python-ironicclient master: Zuul: Remove project name https://review.openstack.org/539741 | 00:11 |
*** rbudden has quit IRC | 00:17 | |
*** r-daneel has quit IRC | 00:18 | |
*** milan has joined #openstack-ironic | 00:32 | |
rloo | TheJulia: yeah, someone 'just' needs to go look at the code and see where we can *not* lock. | 00:35 |
rloo | TheJulia: thx for taking the ActionItem to talk to sambetts|afk about the vif_detach tomorrow. | 00:36 |
*** yangyapeng has quit IRC | 00:36 | |
*** tiendc has joined #openstack-ironic | 01:01 | |
*** jistr has quit IRC | 01:01 | |
*** jistr has joined #openstack-ironic | 01:02 | |
*** lbragstad has quit IRC | 01:10 | |
*** rloo has quit IRC | 01:12 | |
*** yangyapeng has joined #openstack-ironic | 01:19 | |
*** namnh has joined #openstack-ironic | 01:28 | |
*** lbragstad has joined #openstack-ironic | 01:31 | |
*** tuanla____ has joined #openstack-ironic | 01:46 | |
*** milan has quit IRC | 01:46 | |
*** gcb has joined #openstack-ironic | 02:16 | |
*** threestrands has joined #openstack-ironic | 02:27 | |
*** threestrands has quit IRC | 02:27 | |
*** threestrands has joined #openstack-ironic | 02:27 | |
* devananda wonders if l2gw would be needed, if using the arista plugin | 02:35 | |
*** priteau has joined #openstack-ironic | 02:57 | |
*** priteau has quit IRC | 02:58 | |
devananda | still experimenting with ovs ... when I put a node into cleaning, I get a vif and vnic_type: baremetal as expected ... details here - http://paste.openstack.org/show/cM036JeYrmCzGZGGZu0N/ | 03:01 |
devananda | in another screen I'm watching the output of "ovsdb-client monitor Open_vSwitch Port name", which, I _think_, would show any changes that happen when a Port gets plumbed through either virtual or physical networks ... but nothing happens | 03:02 |
*** rbudden has joined #openstack-ironic | 03:08 | |
*** r-daneel has joined #openstack-ironic | 03:10 | |
*** zhangfei has joined #openstack-ironic | 03:10 | |
*** r-daneel has quit IRC | 03:15 | |
*** tiendc has quit IRC | 03:19 | |
*** tuanla____ has quit IRC | 03:19 | |
*** tuanla____ has joined #openstack-ironic | 03:20 | |
*** tiendc has joined #openstack-ironic | 03:20 | |
openstackgerrit | lei zhang proposed openstack/bifrost master: Zuul: Remove project name https://review.openstack.org/539799 | 03:57 |
*** dsneddon has quit IRC | 04:06 | |
*** dsneddon has joined #openstack-ironic | 04:07 | |
*** harlowja has quit IRC | 04:12 | |
*** dsneddon has quit IRC | 04:17 | |
*** crushil has joined #openstack-ironic | 04:20 | |
*** rbudden has quit IRC | 04:20 | |
*** dsneddon has joined #openstack-ironic | 04:26 | |
*** hieulq_ has quit IRC | 04:30 | |
*** links has joined #openstack-ironic | 04:32 | |
*** hieulq_ has joined #openstack-ironic | 04:32 | |
openstackgerrit | Merged openstack/ironic stable/pike: Add no-vendor interface to the idrac hardware types https://review.openstack.org/520010 | 04:33 |
hshiina | hi ironic | 04:41 |
*** jaganathan has joined #openstack-ironic | 04:41 | |
*** harlowja has joined #openstack-ironic | 04:41 | |
hshiina | now, resuce is tested at: https://review.openstack.org/#/c/528704/ | 04:41 |
patchbot | patch 528704 - ironic - Update a CI job for rescue mode | 04:41 |
hshiina | the job to test rescue is ironic-tempest-dsvm-ipa-wholedisk-bios-agent_ipmitool-tinyipa | 04:41 |
hshiina | in test_baremetal_server_ops | 04:42 |
hshiina | but, the test failed: http://logs.openstack.org/04/528704/13/check/ironic-tempest-dsvm-ipa-wholedisk-bios-agent_ipmitool-tinyipa/dd7ace3/logs/screen-ir-api.txt.gz#_Jan_31_13_15_03_655896 | 04:42 |
hshiina | agent_rescue.validate failed. | 04:42 |
hshiina | it seems that reflection.get_signature doesn't work well: https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/agent.py#L790 | 04:43 |
hshiina | this may be caused by the METRIC.timer: https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/pxe.py#L471 | 04:43 |
hshiina | does anyone know how to see the signature of a decorated method? | 04:44 |
*** etingof has quit IRC | 04:49 | |
jlvillal | hshiina, My gut feeling is that it is not the decorator. If it was then likely other tests would have failed. | 05:00 |
*** zhangfei has quit IRC | 05:03 | |
hshiina | jlvillal: thanks, i will look into it more. | 05:04 |
*** etingof has joined #openstack-ironic | 05:11 | |
*** bfournie has quit IRC | 05:29 | |
*** crushil has quit IRC | 05:35 | |
*** dpawar has joined #openstack-ironic | 05:41 | |
*** threestrands has quit IRC | 05:48 | |
*** dsneddon has quit IRC | 05:51 | |
openstackgerrit | Hironori Shiina proposed openstack/ironic master: Add a unit test for AgentRescue.validate() with PXEBoot https://review.openstack.org/539822 | 05:57 |
*** harlowja has quit IRC | 06:12 | |
*** liuzz_ has joined #openstack-ironic | 06:15 | |
*** liuzz has quit IRC | 06:15 | |
*** liuzz_ has quit IRC | 06:32 | |
*** liuzz has joined #openstack-ironic | 06:32 | |
*** liuzz has quit IRC | 06:34 | |
*** zhangfei has joined #openstack-ironic | 06:43 | |
*** dsneddon has joined #openstack-ironic | 06:58 | |
*** mjura has joined #openstack-ironic | 07:04 | |
*** e0ne has joined #openstack-ironic | 07:07 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic master: Updated from global requirements https://review.openstack.org/538463 | 07:10 |
*** liuzz has joined #openstack-ironic | 07:11 | |
*** devananda has quit IRC | 07:13 | |
*** etingof has quit IRC | 07:16 | |
*** etingof has joined #openstack-ironic | 07:17 | |
*** AlexeyAbashkin has joined #openstack-ironic | 07:18 | |
*** marios has joined #openstack-ironic | 07:18 | |
*** vsaienk0 has joined #openstack-ironic | 07:22 | |
sapd_ | hi everyone. I'm setting raid config for the node, my command is "openstack baremetal node set --target-raid-config '{"logical_disks": [ { "size_gb": "MAX", "raid_level": "1", "is_root_volume": true }]}' svr450r", but the output is "Please specify what to set." How to write correct command? help me! | 07:23 |
*** rcernin has quit IRC | 07:23 | |
*** dsneddon has quit IRC | 07:25 | |
*** rbartal has joined #openstack-ironic | 07:30 | |
*** tuanla____ has quit IRC | 07:33 | |
*** namnh has quit IRC | 07:33 | |
*** charcol-v2 has quit IRC | 07:33 | |
*** tuanla____ has joined #openstack-ironic | 07:33 | |
*** namnh has joined #openstack-ironic | 07:33 | |
*** dsneddon has joined #openstack-ironic | 07:36 | |
*** AlexeyAbashkin has quit IRC | 07:37 | |
*** dsneddon has quit IRC | 07:41 | |
*** pcaruana has joined #openstack-ironic | 07:47 | |
*** zhangfei has quit IRC | 07:48 | |
*** AlexeyAbashkin has joined #openstack-ironic | 07:50 | |
hshiina | sapd_: you may hit a bug: https://bugs.launchpad.net/python-ironicclient/+bug/1702120 | 07:55 |
openstack | Launchpad bug 1702120 in python-ironicclient "Wrong warning when setting target_raid_config with OSC plugin" [Medium,Fix released] - Assigned to Kyrylo Romanenko (kromanenko) | 07:55 |
hshiina | please see the version of ironicclient | 07:55 |
*** dsneddon has joined #openstack-ironic | 07:56 | |
openstackgerrit | Dao Cong Tien proposed openstack/ironic master: Update a CI job for rescue mode https://review.openstack.org/528704 | 08:02 |
openstackgerrit | Dao Cong Tien proposed openstack/ironic-tempest-plugin master: Rescue mode standalone tests https://review.openstack.org/538119 | 08:06 |
*** dsneddon has quit IRC | 08:13 | |
*** dsneddon has joined #openstack-ironic | 08:16 | |
*** racedo has joined #openstack-ironic | 08:18 | |
sapd_ | hshiina: I use ironic command line and It worked. | 08:19 |
hshiina | sapd_: i got it | 08:23 |
openstackgerrit | Dao Cong Tien proposed openstack/ironic-tempest-plugin master: Rescue mode tempest tests https://review.openstack.org/528699 | 08:24 |
*** yangyapeng has quit IRC | 08:25 | |
*** yangyapeng has joined #openstack-ironic | 08:26 | |
*** dsneddon has quit IRC | 08:26 | |
*** d0ugal has joined #openstack-ironic | 08:27 | |
*** yangyapeng has quit IRC | 08:31 | |
*** liuzz_ has joined #openstack-ironic | 08:31 | |
*** liuzz has quit IRC | 08:32 | |
*** dsneddon has joined #openstack-ironic | 08:34 | |
*** dsneddon has quit IRC | 08:40 | |
*** Tengu has joined #openstack-ironic | 08:40 | |
Tengu | hello people :). | 08:40 |
Tengu | small question: using VirtualBMC, is it possible for ironic to introspect virtual machines? Apparently, I'm either loacking something in my domain configuration, or there's a small issue… | 08:41 |
Tengu | it seems net-boot doesn't properly work. and this is of course a "small" issue for the whole introspection process, not to mention install/deploy -.-' | 08:42 |
*** dsneddon has joined #openstack-ironic | 08:43 | |
*** dsneddon has quit IRC | 08:48 | |
*** dsneddon has joined #openstack-ironic | 08:54 | |
*** namnh_ has joined #openstack-ironic | 08:58 | |
*** dsneddon has quit IRC | 08:59 | |
*** dsneddon has joined #openstack-ironic | 09:00 | |
*** namnh has quit IRC | 09:02 | |
*** dsneddon has quit IRC | 09:07 | |
openstackgerrit | paresh sao proposed openstack/ironic master: [docs] Firmware based boot from volume for iLO drivers https://review.openstack.org/539158 | 09:11 |
openstackgerrit | paresh sao proposed openstack/ironic master: [docs] Firmware based boot from volume for iLO drivers https://review.openstack.org/539158 | 09:12 |
*** dsneddon has joined #openstack-ironic | 09:14 | |
*** dsneddon has quit IRC | 09:18 | |
*** pcaruana has quit IRC | 09:19 | |
*** pcaruana has joined #openstack-ironic | 09:19 | |
*** athomas has joined #openstack-ironic | 09:20 | |
*** derekh has joined #openstack-ironic | 09:21 | |
*** exodusftw has quit IRC | 09:29 | |
*** MattMan has quit IRC | 09:30 | |
*** MattMan has joined #openstack-ironic | 09:31 | |
*** exodusftw has joined #openstack-ironic | 09:36 | |
*** vsaienk0 has quit IRC | 09:38 | |
*** amarao has joined #openstack-ironic | 09:38 | |
*** milan_ has joined #openstack-ironic | 09:43 | |
*** vsaienk0 has joined #openstack-ironic | 09:51 | |
pas-ha | Tengu: AFAIK inspector has its integration testing running on vbmc-managed VMs, so presumably this should work | 09:53 |
*** priteau has joined #openstack-ironic | 09:57 | |
Tengu | pas-ha: hmm ok. I must be missing something somewhere then. | 09:59 |
Tengu | I could goa bit further, the VMs actually seem to boot with the PXE. but they don't report to ironic. | 09:59 |
Tengu | might be something with the network, still digging. | 09:59 |
*** milan_ has quit IRC | 10:01 | |
*** amarao1 has joined #openstack-ironic | 10:03 | |
*** dsneddon has joined #openstack-ironic | 10:13 | |
*** stendulker has joined #openstack-ironic | 10:15 | |
*** dsneddon has quit IRC | 10:18 | |
*** dsneddon has joined #openstack-ironic | 10:21 | |
openstackgerrit | Merged openstack/ironic master: Add missing ilo vendor to the ilo hardware types https://review.openstack.org/539582 | 10:26 |
*** dsneddon has quit IRC | 10:26 | |
*** AlexeyAbashkin has quit IRC | 10:28 | |
*** priteau has quit IRC | 10:30 | |
*** tuanla____ has quit IRC | 10:37 | |
*** sambetts|afk is now known as sambetts | 10:41 | |
sambetts | Morning all | 10:41 |
etingof | o/ | 10:43 |
*** dsneddon has joined #openstack-ironic | 10:46 | |
hjensas | Good morning o/ | 10:50 |
*** dsneddon has quit IRC | 10:51 | |
*** AlexeyAbashkin has joined #openstack-ironic | 10:53 | |
*** dsneddon has joined #openstack-ironic | 10:54 | |
*** lucas-afk is now known as lucasagomes | 10:56 | |
*** dsneddon has quit IRC | 10:59 | |
*** priteau has joined #openstack-ironic | 11:00 | |
*** olivierbourdon has joined #openstack-ironic | 11:02 | |
*** priteau has quit IRC | 11:04 | |
*** priteau has joined #openstack-ironic | 11:04 | |
openstackgerrit | Merged openstack/ironic master: Flat networks use node.uuid when binding ports. https://review.openstack.org/536040 | 11:07 |
*** mtreinish_ has joined #openstack-ironic | 11:08 | |
*** priteau has quit IRC | 11:09 | |
openstackgerrit | Merged openstack/ironic master: Deprecate classic drivers https://review.openstack.org/536928 | 11:09 |
*** mtreinish has quit IRC | 11:11 | |
*** sdake has quit IRC | 11:11 | |
*** hwoarang has quit IRC | 11:11 | |
*** ianw has quit IRC | 11:11 | |
*** TheJulia has quit IRC | 11:11 | |
*** ildikov has quit IRC | 11:11 | |
*** mtreinish_ is now known as mtreinish | 11:11 | |
*** sdake has joined #openstack-ironic | 11:11 | |
*** sdake has quit IRC | 11:11 | |
*** sdake has joined #openstack-ironic | 11:11 | |
*** homeski has quit IRC | 11:12 | |
*** ildikov has joined #openstack-ironic | 11:12 | |
*** hwoarang has joined #openstack-ironic | 11:12 | |
*** priteau has joined #openstack-ironic | 11:12 | |
*** ianw has joined #openstack-ironic | 11:12 | |
*** homeski has joined #openstack-ironic | 11:12 | |
*** TheJulia has joined #openstack-ironic | 11:12 | |
openstackgerrit | James E. Blair proposed openstack/sushy-tools master: Zuul: Remove project name https://review.openstack.org/539906 | 11:17 |
*** exodusftw has quit IRC | 11:18 | |
*** exodusftw has joined #openstack-ironic | 11:24 | |
*** gcb has quit IRC | 11:25 | |
*** gcb has joined #openstack-ironic | 11:25 | |
*** tiendc has quit IRC | 11:30 | |
*** namnh_ has quit IRC | 11:32 | |
*** AlexeyAbashkin has quit IRC | 11:33 | |
*** AlexeyAbashkin has joined #openstack-ironic | 11:33 | |
*** stendulker has quit IRC | 11:34 | |
openstackgerrit | Derek Higgins proposed openstack/ironic master: Add option to specify mac adress in devstack/.../create-node.sh https://review.openstack.org/539488 | 11:49 |
openstackgerrit | Hironori Shiina proposed openstack/ironic-lib master: DNM: Use six.wraps in metrics.TIMER https://review.openstack.org/539918 | 11:52 |
openstackgerrit | Hironori Shiina proposed openstack/ironic master: Add a unit test for AgentRescue.validate() with PXEBoot https://review.openstack.org/539822 | 11:54 |
*** etingof has quit IRC | 12:09 | |
*** vabada has quit IRC | 12:11 | |
*** vabada has joined #openstack-ironic | 12:12 | |
*** dsneddon has joined #openstack-ironic | 12:13 | |
*** dtantsur|afk is now known as dtantsur | 12:18 | |
dtantsur | morning/afternoon ironic | 12:18 |
*** dsneddon has quit IRC | 12:18 | |
*** stendulker has joined #openstack-ironic | 12:18 | |
dtantsur | whhaaa, what a nostalgic feeling: devananda, BadCub and JayF in channel yesterday (and jroll hanging around again) :) | 12:20 |
openstackgerrit | Hironori Shiina proposed openstack/ironic master: Update a CI job for rescue mode https://review.openstack.org/528704 | 12:20 |
*** etingof has joined #openstack-ironic | 12:22 | |
*** links has quit IRC | 12:28 | |
*** stendulker has quit IRC | 12:30 | |
openstackgerrit | Markos Chandras (hwoarang) proposed openstack/bifrost master: roles: bifrost-create-vm-nodes: Set NIC model to 'virtio' https://review.openstack.org/539258 | 12:30 |
openstackgerrit | Markos Chandras (hwoarang) proposed openstack/bifrost master: roles: bifrost-ironic-install: Download the iPXE loader file https://review.openstack.org/539298 | 12:30 |
*** AlexeyAbashkin has quit IRC | 12:35 | |
*** jlabarre has joined #openstack-ironic | 12:36 | |
openstackgerrit | Sam Betts proposed openstack/ironic master: Only set default network interface flat if enabled in config https://review.openstack.org/535830 | 12:37 |
*** links has joined #openstack-ironic | 12:39 | |
*** amarao has quit IRC | 12:40 | |
*** amarao has joined #openstack-ironic | 12:40 | |
jroll | dtantsur: right? weird eh | 12:41 |
jroll | morning | 12:41 |
*** makowals has quit IRC | 12:43 | |
*** makowals has joined #openstack-ironic | 12:54 | |
*** dprince has joined #openstack-ironic | 12:54 | |
*** makowals has quit IRC | 12:55 | |
etingof | ironic design question: can drivers modify `node.instance_info` fields or `node.instance_info` is only intended for ironic clients to update (e.g. nova)? | 12:57 |
jroll | >>> reflection.get_signature(pxe.PXEBoot.prepare_ramdisk).parameters | 12:59 |
jroll | OrderedDict([('args', <Parameter at 0x2b02a99f6998 'args'>), ('kwargs', <Parameter at 0x41e01b0 'kwargs'>)]) | 12:59 |
jroll | hshiina: I'd say the decorator is influencing the reflection module, yes ^ | 12:59 |
jroll | etingof: I think that's fine, note that it will be blown away when the instance is deleted | 12:59 |
sambetts | etingof: instance_info should be an input only field I believe, if drivers need to store anything then they can use driver_internal_info | 12:59 |
etingof | hmm, conflicting opinions ;) | 13:00 |
jroll | pretty sure some drivers use it | 13:00 |
* jroll could be wrong | 13:00 | |
etingof | iLO does that | 13:00 |
jroll | yeah, that's the one I was thinking of | 13:01 |
sambetts | they do to horribly hack the boot mode | 13:01 |
etingof | hey, that's what I a working on as well - the boot mode | 13:01 |
jroll | I don't have a problem with that, personally, but I don't have a +2 either :P | 13:01 |
etingof | so my code may discover that the bm is configured to do uefi boot while ironic node is configured to do bios boot | 13:02 |
hshiina | jroll: thanks. it seems that that this happens because metrics.TIMER doesn't use six.wraps. | 13:02 |
etingof | I'd like to update ironic node info to make it doing uefi boot | 13:02 |
etingof | so ironic and bm nodes boot modes would be consistent | 13:03 |
sambetts | your node should have its properties set correctly to only support UEFI if you care about which mode boot your using | 13:03 |
sambetts | then the flavor will pass the right boot mode down to the ndoe | 13:03 |
*** links has quit IRC | 13:03 | |
sambetts | s/flavor/nova/ | 13:04 |
jroll | hshiina: possibly, not sure | 13:04 |
dtantsur | morning jroll :) | 13:04 |
*** rbartal has quit IRC | 13:04 | |
dtantsur | etingof: I think I'm on "better don't" side of the argument :) | 13:04 |
etingof | aha, so I should never sync actual bm boot node to ironic? but only the other way - ironic node -> bm | 13:05 |
*** zhangfei has joined #openstack-ironic | 13:05 | |
etingof | I am referring to the boot mode setting here | 13:05 |
dtantsur | etingof: ironic should be aware of the actual boot mode, but you don't have to store it | 13:05 |
dtantsur | maybe in the internal_driver_info, if you need it for any reason | 13:05 |
dtantsur | but you don't have to modify capabilities, for example | 13:05 |
dtantsur | (except during inspection, which is not related to the question here) | 13:06 |
etingof | once I discover actual bm boot mode, I need to convey it to the rest of ironic internals | 13:06 |
*** weshay|ruck is now known as weshay | 13:06 | |
dtantsur | etingof: I think it's the other way around: we need to modify the deploy procedure to ask for it | 13:07 |
etingof | easy way is to update node.instance_info because other parts of ironic would pick it up from there | 13:07 |
*** serlex has joined #openstack-ironic | 13:08 | |
dtantsur | e.g. if the PXE boot cares (it does), it should call something like get_requested_boot_mode, which falls back to task.driver.management.get_current_boot_mode | 13:08 |
dtantsur | (all names are random, I'm not sure) | 13:08 |
sambetts | the key is an instance that wants bios should never be scheduled onto a node that is only uefi | 13:09 |
sambetts | if the node can change it then advertise both mode for that node and then change it at deploy time | 13:10 |
etingof | yes, this is what is happening. now, how do we pass the discovered bm boot mode to other functions that change their behavior depending on node.instance_info? | 13:10 |
*** jcoufal has joined #openstack-ironic | 13:10 | |
dtantsur | etingof: we have to modify these "other functions" to call something instead of relying on instance_info only | 13:11 |
sambetts | etingof: if you node is only UEFI and can't change it, then it should onyl have UEFI it its capabilites | 13:11 |
etingof | dtantsur, aha, `driver_internal_info` perhaps | 13:11 |
dtantsur | lemme draft something | 13:12 |
sambetts | we have this function already https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/deploy_utils.py#L761 | 13:12 |
sambetts | it gets the capability of the node first | 13:13 |
sambetts | that overrides everything | 13:13 |
*** AlexeyAbashkin has joined #openstack-ironic | 13:13 | |
sambetts | then it uses instance_info | 13:14 |
*** trown|outtypewww is now known as trown | 13:14 | |
sambetts | which means if a node is capable of only one type of boot mode, then operator sets the capabilty to the right one (no sync, although maybe it should be an inspection step) | 13:15 |
dtantsur | etingof, sambetts, I'm thinking of something like https://etherpad.openstack.org/p/ironic-detect-boot-mode | 13:15 |
sambetts | if the node is capable of either, then it uses which ever is either already configured on the server if not specified, other wise uses it from instance_info | 13:15 |
jroll | hshiina: I really don't think that we should be inspecting method parameters here. that means we've failed somewhere at making a good boot interface. I'm going to look and see if there's a better way to do this. would you file a bug when you have time, please? | 13:15 |
etingof | sambetts, dtantsur here's the logic I have in mind -- https://git.openstack.org/cgit/openstack/ironic/tree/ironic/drivers/modules/pxe.py?h=refs/changes/73/526773/9#n322 | 13:16 |
hshiina | jroll: sure. i'm writing a report. | 13:16 |
jroll | hshiina: thank you :) | 13:17 |
dtantsur | etingof: it's quite correct, but the caching should be in driver_internal_info, if we even need it | 13:17 |
etingof | dtantsur, that implies that I need to look into driver_internal_info *in addition* to instance_info and capabilities | 13:18 |
sambetts | the idea is to have ironic make sure the capability specified by the admin is correct | 13:18 |
sambetts | ? | 13:18 |
*** yangyapeng has joined #openstack-ironic | 13:19 | |
etingof | sambetts, my current approach is to try to make boot modes consistent if possible so the instance would boot | 13:19 |
dtantsur | etingof: it's going to be one function in the end, which you just call whenever needed | 13:20 |
etingof | dtantsur, it's going to be at least this -- https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/deploy_utils.py#L761 | 13:21 |
sambetts | what I'm saying is that today the capabilities field and the physical server should be in sync manually that is the contract | 13:21 |
sambetts | with the operator | 13:21 |
sambetts | if they aren't then ironic is misconfigured | 13:21 |
sambetts | and thats a operator error | 13:21 |
dtantsur | etingof: see updated etherpad | 13:22 |
dtantsur | good point sambetts, yeah | 13:22 |
etingof | sambetts, right, now the question is -- do we want some more magic as explained here? -- https://git.openstack.org/cgit/openstack/ironic/tree/ironic/drivers/modules/pxe.py?h=refs/changes/73/526773/9#n325 | 13:22 |
*** dpawar has quit IRC | 13:23 | |
dtantsur | etingof: updated the etherpad with sambetts's point | 13:24 |
sambetts | I'm not sure we need it, because the way it is today, if you node can't have its boot mode set dynamically, and the operator has configured ironic correctly, then ironic will use the mode it knows which is the correct one, otherwise either the instance doesn't care or ironic will try to dynamically set it | 13:26 |
sambetts | we just need to be very clear and state "if your node can't change its boot mode dynamically then you must inform ironic using the node's capabilites field which boot mode it supports" | 13:28 |
sambetts | otherwise you run into failing deployments because an admin has misconfigured ironic's boot mode information | 13:29 |
dtantsur | ++ I added it to line 17 | 13:30 |
dtantsur | (and yes, we have it very often in tripleo) | 13:30 |
nicodemos | good morning, ironic | 13:30 |
sambetts | dtantsur: at L17 that fails deployments, if we just insist that operators configure the capabilties correct in the first place then we don't blow up deployments for operator errors | 13:31 |
etingof | sambetts, so the alternative to the failure at #17 is to set ironic boot mode to bm boot mode | 13:32 |
etingof | sambetts, what I understand you think is not good ;) | 13:32 |
dtantsur | sambetts: how do we insist? | 13:32 |
dtantsur | you mean, update validate() with this check? | 13:33 |
sambetts | possibly if we want actively enforce it, but we don't actively inforce anything else the operator informs us about | 13:34 |
sambetts | s/inforce/enforce | 13:34 |
etingof | btw, validate() may be more racy if bm node can change its boot mode in the meantime | 13:35 |
etingof | between validation and deployment | 13:35 |
*** SerenaFeng has joined #openstack-ironic | 13:35 | |
hshiina | jroll: i filed a bug: https://bugs.launchpad.net/ironic/+bug/1746730 | 13:35 |
openstack | Launchpad bug 1746730 in Ironic "AgentRescue validation fails to get the signature of PXEBoot.prepare_ramdisk" [Undecided,New] | 13:35 |
dtantsur | sambetts: what exactly do you have in mind? I can only think about validate. as etingof mentions it's potentially racy, but we can live with it. | 13:35 |
sambetts | which puts it in sync power state teritory | 13:35 |
jroll | hshiina: thank you, I'm looking into it already | 13:35 |
hshiina | jroll: thank you | 13:36 |
jroll | hshiina: if you have ideas I'd love to hear them :) | 13:37 |
*** dsneddon has joined #openstack-ironic | 13:38 | |
sambetts | dtantsur, etingof: I think first we should make it very clear in the docs that if a node is unable to set its boot mode then capabilities should be set to just one mode | 13:39 |
dtantsur | ++ | 13:39 |
sambetts | if capabilties are set then that prevents instances requiring a different boot mode from being scheduled onto that node | 13:39 |
*** makowals has joined #openstack-ironic | 13:40 | |
sambetts | if capabilties are not set then we have two forks, if instance_info has a boot mode set, we try to set that boot mode if we succeed great, if we fail, deploy fails. | 13:41 |
sambetts | if capabilties are not set and instance_info is not set, then we call a management.get_boot_mode function and use that | 13:41 |
sambetts | otherwise fallback to bios | 13:42 |
sambetts | or a config file default | 13:43 |
etingof | sambetts, that is clear, but did you mention the condition at line #17? do we fail if apparent/dynamic bm boot mode is different from capabilities? | 13:44 |
hshiina | jroll: i once propsed another way with interface capabilities in reviewing: https://review.openstack.org/#/c/400437/83/ironic/drivers/modules/agent.py@797 | 13:45 |
patchbot | patch 400437 - ironic - Agent rescue implementation (MERGED) | 13:45 |
*** dsneddon has quit IRC | 13:46 | |
sambetts | etingof: we don't check, if the operator has told us this server is boot mode uefi only then believe them and use that, the alternative is to have a periodic check which checks the servers boot mode like sync power state | 13:47 |
sambetts | and that will enforce it if it can, if it can't then it maintaince mode's the node | 13:47 |
jroll | hshiina: yeah, that is one way I was thinking of, the other is just assuming the boot interface can do rescue | 13:47 |
etingof | sambetts, can we instead of the periodic check go out and check the current bm boot mode right at the time of deployment? if it is different from capabilities - change bm boot mode | 13:49 |
*** makowals has quit IRC | 13:53 | |
sambetts | etingof: we shouldn't change the boot mode if the operator has told us to run this server only in UEFI mode for example | 13:53 |
sambetts | so we should fail | 13:53 |
etingof | alright, thank you sambetts ad dtantsur | 13:54 |
sambetts | and if we fail at deployment time, then a bad opertor config is failing a deployment for a user | 13:54 |
sambetts | etingof: btw because this touches capabilties a lot of this is tied up in the ongoing traits work to clean caps up | 13:55 |
etingof | I see | 13:56 |
*** trown is now known as trown|brb | 13:56 | |
*** Goneri has joined #openstack-ironic | 13:56 | |
sambetts | etingof: i think dtantsur had some thoughts on dynamic traits which might be the right place for an automagically updating boot mode trait | 13:56 |
*** bfournie has joined #openstack-ironic | 13:56 | |
*** SerenaFeng has quit IRC | 13:56 | |
*** trown|brb is now known as trown | 13:58 | |
*** crushil has joined #openstack-ironic | 13:59 | |
openstackgerrit | Pavlo Shchelokovskyy proposed openstack/ironic master: Do not use asserts with business logic https://review.openstack.org/538991 | 13:59 |
*** makowals has joined #openstack-ironic | 13:59 | |
*** cdearborn has joined #openstack-ironic | 14:00 | |
hshiina | jroll: thanks. I'm afraid i have to leave the office. | 14:02 |
jroll | hshiina: no problem, have a good day :) | 14:02 |
hshiina | it's night here | 14:02 |
hshiina | good night | 14:02 |
*** hshiina is now known as hshiina|afk | 14:02 | |
*** rloo has joined #openstack-ironic | 14:05 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic master: Remove mode argument from boot.(prepare|clean_up)_ramdisk https://review.openstack.org/539961 | 14:06 |
jroll | dtantsur: curious if you're okay with just dropping that or if I should deprecate it properly ^ | 14:06 |
jroll | it's been on master for six days, I know what the right thing to do is, not sure if it's worth it :) | 14:07 |
* jroll still needs to fix tests | 14:08 | |
*** saop has joined #openstack-ironic | 14:08 | |
saop | TheJulia, Hello | 14:09 |
rloo | jroll: i think it is ok to just remove it. | 14:09 |
TheJulia | good morning everyone | 14:09 |
* TheJulia was trying to tip toe into the room this morning | 14:09 | |
TheJulia | and saop caught me :) | 14:09 |
* rloo wonders why we want to remove it, will take a look | 14:10 | |
jroll | morning rloo, TheJulia :) | 14:10 |
rloo | good morning TheJulia, jroll, dtantsur, saop, and everyone else! | 14:10 |
jroll | rloo: yeah, when you have time to review the commit/bug | 14:10 |
jroll | no huge rush, but rescue is totally broken :) | 14:10 |
saop | TheJulia, Can you please review this patch https://review.openstack.org/#/c/539158/ | 14:11 |
patchbot | patch 539158 - ironic - [docs] Firmware based boot from volume for iLO dri... | 14:11 |
rloo | jroll: ugh. that was my idea too. but i thought it had been tested :-( | 14:12 |
jroll | rloo: haven't looked yet, but likely tested with a dummy boot interface that has no decorators | 14:12 |
*** jhesketh has quit IRC | 14:12 | |
rloo | jroll: well, i think it was my idea. i did mention passing mode to something... :-( | 14:13 |
rloo | jroll: hmm, well, that isn't really testing. sigh. i hate that we are rushed. | 14:13 |
*** jhesketh has joined #openstack-ironic | 14:13 | |
TheJulia | dtantsur: I know, so nostalgic last night :) | 14:13 |
jroll | rloo: it was yours, not your fault for not realizing it wouldn't work | 14:13 |
*** makowals has quit IRC | 14:13 | |
dtantsur | jroll: I guess I'm fine | 14:14 |
dtantsur | TheJulia: :) | 14:14 |
dtantsur | morning rloo, TheJulia | 14:14 |
*** saop has quit IRC | 14:14 | |
jroll | dtantsur: thanks for the +2, need to update tests first though :P | 14:14 |
* jroll hackhackhack | 14:14 | |
dtantsur | :) | 14:15 |
*** rbudden has joined #openstack-ironic | 14:15 | |
TheJulia | sambetts: by chance did you see the discussion yesterday regarding vif_attach? | 14:15 |
sambetts | TheJulia: yeah, I read through what I could see in the backscroll | 14:15 |
* dtantsur -1 the ilo rescue patch | 14:16 | |
*** makowals has joined #openstack-ironic | 14:16 | |
jroll | huh, no release note for the boot interface change, less work for me :o | 14:17 |
sambetts | TheJulia: I think the difficulty I've been having is that in nova for example when you delete an instance it actually deletes the instance record and therefore you expect everthing associated with it to be deleted | 14:18 |
openstackgerrit | Merged openstack/ironic-python-agent stable/pike: Zuul: Remove project name https://review.openstack.org/538845 | 14:18 |
sambetts | TheJulia: however in ironic when we delete we're just changing the state on that server, so I'm not sure where to draw the line, part of me thinks that we should remove all instance related information from the node, the other part of me thinks we should undeploy the node but leave all the user configs set on the baremetal node because then they could move it straight back to active to redeploy | 14:20 |
TheJulia | sambetts: true, but we are also racey in that we are super locky on that entire sequence of events after the node is set to a deleted provision state. The consensus seems to be, for us to at least do the same as we do with instance_info and volume_targets to vif attachment records. | 14:20 |
sambetts | the server | 14:20 |
TheJulia | sambetts: so the contract, and even only way to keep it working short of removing all of our super heavy locking is to clean up upon tear_down by removing instance related information locally. Nova could still try to delete and do the right thing to verify, but should stop when it sees the items are gone. | 14:22 |
TheJulia | I think we could add an option to make an operator be able to choose the alternative behavior of leave all instance info in place, but... | 14:22 |
TheJulia | then there be dragons kind of warnings will be needed around that option since it won't play well with nova at all | 14:22 |
*** rbartal has joined #openstack-ironic | 14:23 | |
sambetts | yeah I think I'm more strongly leant towards fixing volmes and instance_info to not delete them either, from personal experince operating standalone the fact that ironic nukes instance_info is super irritating | 14:23 |
sambetts | because without something like bifrost to repopulate it, if I want to delete and clean then redeploy the same way I had before, I have to manually set all the data again | 14:25 |
* jroll has a thought... PUT /v1/nodes/foo/states/provision {'state': 'tear_down', 'clean_instance_config': true} | 14:25 | |
TheJulia | oh god, that just pushes this issue into late next cycle then | 14:26 |
rloo | jroll: wrt 539961, don't you want to remove the 'mode' parameter from BootInterface.clean_up_ramdisk()? | 14:26 |
TheJulia | sambetts: what is wrong with rebuild if your tearing down and redeploying? | 14:26 |
jroll | rloo: oh yeah, need to hit the base class, thanks | 14:27 |
TheJulia | I guess a lack of cleaning in that | 14:27 |
rloo | sambetts has a point, if instance_info is something that the user sets, we shouldn't be clearing it internally. but we do add stuff to it :-( | 14:27 |
*** alex_xu has quit IRC | 14:27 | |
sambetts | yeah, theres a lack of cleaning, and also you might not want to redeploy imediately | 14:27 |
jroll | TheJulia: not advocating, just putting all options on the table | 14:27 |
TheJulia | jroll: I actually like the idea of a tear_down state command that just rips the machine down and doesn't delete the fields | 14:28 |
*** jcoufal_ has joined #openstack-ironic | 14:28 | |
TheJulia | s/command/api verb/ | 14:28 |
dtantsur | just keep in mind that ideally we should eventually get rid of instance_info in favor of something restful(ish) | 14:28 |
sambetts | POST /v1/deployments | 14:28 |
sambetts | ;) | 14:28 |
*** rloo has quit IRC | 14:28 | |
TheJulia | with a v2 api? | 14:28 |
jroll | and we've successfully made this into a mountain \o/ | 14:29 |
sambetts | \\o/ | 14:29 |
TheJulia | woot! | 14:29 |
dtantsur | sambetts++ | 14:30 |
*** rloo has joined #openstack-ironic | 14:30 | |
dtantsur | like we have a semi-virtual Introspection object in ironic-inspector, which represents the process itself | 14:30 |
*** alex_xu has joined #openstack-ironic | 14:30 | |
*** jcoufal has quit IRC | 14:31 | |
jroll | TheJulia: I guess I'm back to thinking we probably don't need to lock for vif-detach, but I don't have evidence to support that case in either direction | 14:32 |
*** lucasagomes is now known as lucas-hungry | 14:32 | |
TheJulia | dtantsur: how do you feel about putting a configuration option in that allows for the fundamental logic change in the short term to provide stand-alone users with the ability to have our current teardown process preserve things about the node that was deployed... and we then add vif records to the list of things that should be removed upon tear_down. And maybe we add a verb next cycle that allows more fine | 14:32 |
TheJulia | grained takedown of the node. | 14:32 |
sambetts | jroll: before we added vif detach as an API, nova used to remove the vif_port_ids from the ports during cleaning | 14:33 |
TheJulia | maybe we should make ports lockable.... | 14:33 |
jroll | TheJulia: I think that was actually in a lock, I seem to remember having the same problem a while back | 14:34 |
TheJulia | jroll: huh? | 14:34 |
dtantsur | TheJulia: O | 14:34 |
dtantsur | damn | 14:34 |
jroll | definitely didn't have the vif attach/detach api at the time, but also had downstream hacks, so not 100% sure | 14:34 |
dtantsur | TheJulia: I'm fine with the jroll's idea | 14:34 |
* TheJulia slides a whiskey to dtantsur | 14:34 | |
dtantsur | just pointing out that we should rethink all this long term | 14:35 |
dtantsur | oh, that would not hurt :) | 14:35 |
TheJulia | dtantsur: absolutely agree++ | 14:35 |
sambetts | +1000 | 14:35 |
* TheJulia opens the ptg ideas etherpad | 14:35 | |
*** hshiina|afk has quit IRC | 14:35 | |
jroll | TheJulia: in case you're still huh'ing, before the vif attach/detach thing, we had issues with ports hanging around. but now that I think more, I'm thinking it was a problem with them not being deleted from neutron, so the MAC address was taken. ignore me :) | 14:36 |
sambetts | jroll: that actually reminds me of a bug I hit downstream /me scrambles to check if it was ever fixed | 14:37 |
*** rbartal has quit IRC | 14:37 | |
sambetts | jroll: nope, not fixed... me puts up a patch | 14:40 |
rloo | jroll: ok, i am done with my comments for 539961. for now :) | 14:41 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Migrate the remaining classic drivers to hardware types https://review.openstack.org/539589 | 14:41 |
dtantsur | jroll, rloo ^^^ | 14:41 |
dtantsur | I hope I did not miss anything | 14:42 |
* rloo looks forward to another exciting day of reviews and discussions :D | 14:43 | |
jroll | rloo: thanks, that's helpful | 14:44 |
*** r-daneel has joined #openstack-ironic | 14:47 | |
*** baha has joined #openstack-ironic | 14:48 | |
*** mgoddard_ has joined #openstack-ironic | 14:49 | |
rloo | dtantsur: before i forget, i noticed that the api-ref had reference to classic driver. do you have a list of docs that need updating wrt classic driver deprecation? | 14:49 |
*** AlexeyAbashkin has quit IRC | 14:50 | |
dtantsur | rloo: I don't have a complete list.. I was doing 'grep' the last time | 14:50 |
*** AlexeyAbashkin has joined #openstack-ironic | 14:51 | |
rloo | dtantsur: ok, something to check before end of next week. if i get a chance, i'll take a look today. but for sure, i know api-ref :) https://developer.openstack.org/api-ref/baremetal/#show-driver-properties | 14:52 |
dtantsur | rloo: thanks! in any case, docs fixes are backportable (though we better finish them) | 14:52 |
rloo | dtantsur: right, but i do NOT want to spend time on this stuff after the release. I want to move on :D | 14:53 |
dtantsur | fair enough | 14:53 |
rloo | dtantsur: but we have next week to do docs still ;) | 14:53 |
* rloo thinks it is going to be a BEEEEE-UUUUU-tiful release for Queens :) | 14:54 | |
dtantsur | it is :) | 14:54 |
dtantsur | just to clarify: is the multinode grenade broken again? | 14:55 |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic master: Remove mode argument from boot.(prepare|clean_up)_ramdisk https://review.openstack.org/539961 | 14:56 |
rloo | dtantsur: dunno, haven't looked recently. i hope not... | 14:56 |
TheJulia | I saw it started failing again in the last couple days, but the times I saw it... we had no logs | 14:56 |
dtantsur | ah, right, that's why I did not check | 14:57 |
dtantsur | so the latest failure http://logs.openstack.org/30/535830/2/check/ironic-grenade-dsvm-multinode-multitenant/9ea0d43/ | 14:57 |
*** rbartal has joined #openstack-ironic | 14:57 | |
TheJulia | jroll: ahhh! | 14:58 |
TheJulia | rloo: exciting as always! | 14:58 |
dtantsur | ssh timeout wut http://logs.openstack.org/30/535830/2/check/ironic-grenade-dsvm-multinode-multitenant/9ea0d43/logs/grenade.sh.txt.gz#_2018-02-01_14_40_28_980 | 14:58 |
TheJulia | Okay, I need to eat some breakfast and have more coffee | 14:58 |
jroll | TheJulia: ? | 14:59 |
TheJulia | jroll: in response to the huh context | 14:59 |
jroll | ah | 14:59 |
jroll | :) | 14:59 |
TheJulia | fwiw, I added a huge topic to the ptg ideas etherpad with a number of links | 15:00 |
rloo | dtantsur: network related? (we always like to blame neutron) | 15:01 |
dtantsur | rloo: I cannot completely rule out that it's caused by the patch | 15:01 |
dtantsur | but the regular grenade did not fail https://review.openstack.org/#/c/535830/ | 15:01 |
patchbot | patch 535830 - ironic - Only set default network interface flat if enabled... | 15:01 |
dtantsur | sambetts: looking at the failure on your patch ^^^ | 15:01 |
dtantsur | I see a few PXE booting failures, like http://logs.openstack.org/30/535830/2/check/ironic-grenade-dsvm-multinode-multitenant/9ea0d43/logs/old/ironic-bm-logs/node-1_no_ansi_2018-02-01-14:40:03.txt.gz | 15:01 |
rloo | dtantsur: oh, heh, i didn't even look to see what the patch was. let's keep an eye out... | 15:02 |
dtantsur | okay, I see a few passes green, then red on the oneview patch which is obviously unrelated: http://logs.openstack.org/37/539237/3/check/ironic-grenade-dsvm-multinode-multitenant/52bb6a7/ | 15:04 |
sambetts | dtantsur: its times like this I wish the test logs contained a snapshot of the DB to be able to see which interfaces where actually configured on the nodes | 15:04 |
dtantsur | sambetts: I'd not mind it | 15:04 |
dtantsur | okay, unrelated patch, and we have seemingly the same problems: http://logs.openstack.org/37/539237/3/check/ironic-grenade-dsvm-multinode-multitenant/52bb6a7/logs/old/ironic-bm-logs/node-5_no_ansi_2018-01-31-16:40:30.txt.gz | 15:05 |
sambetts | dtantsur: looking at the logs for my patch it appears to be using the neutron interface as it should be | 15:06 |
*** ijw has joined #openstack-ironic | 15:06 | |
sambetts | and nothing is using the flat interface | 15:06 |
TheJulia | sambetts: some people have updated jobs to preserve databases... just saying :) | 15:06 |
sambetts | TheJulia: oh? I'm surprised we can do that given the limits on the amount we can upload to logs.openstack | 15:07 |
sambetts | that would be awesome if we could do that | 15:07 |
rloo | dtantsur: this one fails too with ssh timeout: http://logs.openstack.org/40/536040/15/check/ironic-grenade-dsvm-multinode-multitenant/cc28c85/logs/grenade.sh.txt.gz#_2018-02-01_08_39_52_689 | 15:07 |
dtantsur | sambetts: hmm, are we supposed to use (i)PXE boot in the multinode grenade at all? | 15:08 |
TheJulia | sambetts: yeah, given last week... I'd say be careful of doing it in the long term since logs.o.o ran out of space last week :( | 15:08 |
dtantsur | I mean, for instances | 15:08 |
sambetts | I think it always uses iPXE for all cases and all jobs, and we have done that for a while now | 15:09 |
sambetts | I think JayF made that change | 15:09 |
sambetts | iirc | 15:09 |
dtantsur | sambetts: I mean, this job is testing networking flipping, no? we cannot PXE-boot instances in this case, can we? | 15:09 |
* dtantsur is confused | 15:10 | |
sambetts | oh... yeah... if its at the point when its booting the tenant image it shouldn't be PXEing | 15:10 |
dtantsur | well, maybe? I'm not sure | 15:10 |
dtantsur | the failed test is about rebooting a server though | 15:10 |
sambetts | there was a case when Ironic didn't set the boot device persistently that it would boot from disk the first time and then pxe if rebooted | 15:12 |
sambetts | but I thought we fixed that | 15:12 |
rloo | dtantsur: according to http://ci-watch.tintri.com/project?project=ironic&time=7+days, multi-grenade started failing and is still failing | 15:12 |
rloo | dtantsur: the first failure is the ssh timeout issue. | 15:13 |
*** olivierbourdon has quit IRC | 15:14 | |
dtantsur | meh.. | 15:15 |
dtantsur | sambetts: do you remember what was the bug and the fix? | 15:15 |
openstackgerrit | Sam Betts proposed openstack/ironic master: Ensure we unbind flat network ports and clear mac addresses https://review.openstack.org/539981 | 15:15 |
sambetts | dtantsur: just look | 15:16 |
sambetts | ing | 15:16 |
dtantsur | thnx | 15:16 |
sambetts | jroll: ^ that fix I've just pushed is the a first interation on fixing the last behind mac addresses thing, putting it up to save it but I havn't had a chance to test it probably yet | 15:17 |
sambetts | iteration* | 15:18 |
* dtantsur has applied for an Irish visa btw | 15:18 | |
sambetts | \o/ | 15:18 |
dtantsur | fingers crossed now :) | 15:19 |
sambetts | dtantsur: ah I was thinking the other way around I guess: https://github.com/openstack/ironic/commit/c7091fb8e2898e526f8ce242f50096a2cabeb1fa | 15:19 |
sambetts | but could be relatd | 15:19 |
*** rpioso|afk is now known as rpioso | 15:20 | |
jroll | sambetts: oh yeah, that'll do it, I think. kind of a backwards way to do it but hey | 15:20 |
sambetts | jroll: I hit the bug when I pre created a port in neutron, and passed it to nova, and of course ironic set the mac address on it, but then after I deleted the instance in nova, it didn't change the mac address back | 15:22 |
sambetts | so the only option you had was to delete the port anyway | 15:22 |
rpioso | Good morning, all | 15:22 |
jroll | sambetts: yep, and anyone else scheduled to that node would get a failure as the mac address was occupied | 15:22 |
sambetts | exactly :( | 15:22 |
jroll | been there, don't remember why the port wasn't getting deleted though | 15:23 |
jroll | neutron failures, I guess | 15:23 |
sambetts | if you weren't manually handing a port to nova, then nova should have deleted it unless something unexpected happened | 15:24 |
jroll | yes, something unexpected definitely happened :P | 15:25 |
sambetts | "OpenStack the place were unexpected things happen" :D | 15:25 |
jroll | sambetts: s/openstack/nexus/ | 15:25 |
* jroll ducks | 15:25 | |
dtantsur | morning rpioso | 15:25 |
rpioso | dtantsur: :) | 15:26 |
sambetts | jroll: ;) *cough* I'm not saying anything | 15:26 |
sambetts | dtantsur: not sure if this is anything but http://logs.openstack.org/37/539237/3/check/ironic-grenade-dsvm-multinode-multitenant/52bb6a7/logs/screen-q-dhcp.txt.gz has a big ol guru meditation at the bottom of it | 15:33 |
dtantsur | sambetts: good find! and I think the 1st failure showed up yesterday, 11:30am my time | 15:34 |
dtantsur | (10:30 UTC?) | 15:34 |
*** zhangfei has quit IRC | 15:34 | |
sambetts | dtantsur: you england +1? | 15:34 |
sambetts | we're on UTC atm | 15:34 |
dtantsur | I should be, yes | 15:34 |
dtantsur | the same meditation in the first failure: http://logs.openstack.org/18/524118/19/check/ironic-grenade-dsvm-multinode-multitenant/00c8e34/logs/screen-q-dhcp.txt.gz#_Jan_31_10_21_06_158548 | 15:36 |
dtantsur | let's check a successful run | 15:36 |
dtantsur | the previous successful run does not have it: http://logs.openstack.org/40/536040/14/check/ironic-grenade-dsvm-multinode-multitenant/8f2a142/logs/screen-q-dhcp.txt.gz | 15:38 |
*** lucas-hungry is now known as lucasagomes | 15:38 | |
* dtantsur goes to #openstack-neutron | 15:39 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic master: Remove mode argument from boot.(prepare|clean_up)_ramdisk https://review.openstack.org/539961 | 15:41 |
*** ijw has quit IRC | 15:41 | |
jlvillal | jroll, I only read part of the scrollback. So you think the decorator is going the introspection to fail? | 15:51 |
jlvillal | jroll, introspection of the method that is. | 15:51 |
jroll | jlvillal: yes, I pasted some python in scrollback that shows it | 15:51 |
jlvillal | jroll, I'm surprised we haven't seen that before. | 15:51 |
jroll | it's new code | 15:51 |
jlvillal | Oh. I thought that code had been in for awhile. Ah | 15:52 |
jroll | 6 days | 15:52 |
jlvillal | Ah! | 15:52 |
jlvillal | Why do I remember introspection code like that being in there a year or more... | 15:52 |
jlvillal | Some other part of the code base I guess | 15:52 |
jroll | jlvillal: there's another example of it, but I'm not sure if that one is wrapped | 15:53 |
jlvillal | jroll, Thanks | 15:53 |
jroll | :) | 15:53 |
jlvillal | Here is a short ~3 hour video on decorators and descriptors ;) https://www.youtube.com/watch?v=81S01c9zytE | 15:54 |
rloo | jlvillal, jroll: it was some new 'timeout' arg, probably worth checking, although I had the feeling it worked... | 15:54 |
rloo | sigh. /me looks... | 15:54 |
jlvillal | rloo, Yeah 'timeout' is what I was thinking of. Thanks. | 15:55 |
jlvillal | rloo, The video suggestion was more of a joke since it is so long. Though I am thinking about watching it at some point. | 15:56 |
rloo | PowerInterface.set_power_state() Probably doesn't work there either, I see @METRICS.time() above at least one of them. | 15:57 |
rloo | jlvillal: you interested in working/fixing that? ^^ | 15:58 |
dtantsur | I think we should now remove the check around "timeout", as we gave people enough time to change their drivers | 15:59 |
dtantsur | but does it mean that it has never worked? :) | 15:59 |
rloo | dtantsur: yup, code doesn't lie. i suspect it never worked. just wondering how it was tested. maybe no one has used timeouts... | 16:00 |
dtantsur | omg | 16:00 |
* dtantsur screw it, I'm done, I'm so done | 16:00 | |
jlvillal | rloo, I think need to fix @METRICS.timer, if we can | 16:00 |
rloo | dtantsur: well, maybe someone should verify that it didn't work :) | 16:00 |
rloo | dtantsur: huh? you can't be done. we're all in this together :) | 16:00 |
jroll | jlvillal: ++, though that will need to be in rocky because freeze | 16:00 |
rloo | well, the way to fix it now is to just make the call and see if an exception is raised, right? | 16:01 |
dtantsur | rloo: kind of flipping tables here :) anyway, we should fix it | 16:01 |
jlvillal | jroll, Is someone already working in that area? | 16:01 |
jroll | rloo: for now, yes | 16:01 |
jroll | jlvillal: not that I know of | 16:01 |
jlvillal | jroll, okay | 16:02 |
jlvillal | I'll take a look later when I get out of bed :) | 16:02 |
rloo | jlvillal: you interested in fixing that? ^^ Otherwise, i can do it. Or anyone else out there in irc-land interested? almost a low-hanging-fruit. | 16:02 |
rloo | jlvillal: ok, thx. we should get that in for the release. i'll take a look to see which/if there are any intree drivers missing that parameter. | 16:02 |
jlvillal | rloo, I'll look. See if six.wraps() helps | 16:03 |
jlvillal | rloo, It is in ironic-lib, that code. | 16:03 |
rloo | jlvillal: please, nothing fancy so late in the game. i mean, just fix the code in ironic, in conductor/utils.py | 16:03 |
jlvillal | METRICS.timer() | 16:03 |
rloo | jlvillal: forget it, i'll fix it. you can do the long term fix. | 16:03 |
jlvillal | rloo, Ah. okay. | 16:03 |
rloo | jlvillal: just that it doesn't matter, if you fix that decorator, what if we have a new decorator... | 16:04 |
dtantsur | rloo, TheJulia, https://review.openstack.org/#/c/539999/ I guess we are expected to do it | 16:04 |
patchbot | patch 539999 - sushy (stable/queens) - Restores sushy session functionality. | 16:04 |
jroll | jlvillal: six.wraps() adds the __wrapped__ attribute, which is used when looking up the signature, so yes, that will do it. https://github.com/aliles/funcsigs/blob/master/funcsigs/__init__.py#L87 | 16:04 |
rloo | dtantsur: oh, a backport already. fun. | 16:04 |
jlvillal | jroll, Thanks | 16:05 |
jroll | np | 16:05 |
TheJulia | dtantsur: ugh, forgot about that one :( | 16:05 |
*** SerenaFeng has joined #openstack-ironic | 16:11 | |
*** lucasagomes has quit IRC | 16:15 | |
*** sauloaislan has joined #openstack-ironic | 16:15 | |
sauloaislan | Morning! | 16:15 |
*** lucasagomes has joined #openstack-ironic | 16:18 | |
sauloaislan | I'm having a problem with Ironic | 16:20 |
sauloaislan | http://paste.openstack.org/show/658328/ | 16:20 |
*** e0ne has quit IRC | 16:22 | |
sauloaislan | In keystoneauth1/seesion.py:982 get 'service_auth' = None | 16:22 |
TheJulia | sauloaislan: what version of keystoneauth1 and what version of ironic? | 16:29 |
sauloaislan | TheJulia: keystoneauth1==3.4.0 | 16:30 |
sauloaislan | ironic==9.2.1.dev89 | 16:31 |
*** crushil has quit IRC | 16:35 | |
NobodyCam | Good Morning Ironic'ers | 16:37 |
dtantsur | morning NobodyCam | 16:37 |
NobodyCam | hey hey dtantsur :) Good morning :) | 16:37 |
TheJulia | sauloaislan: try https://github.com/openstack/requirements/blob/stable/pike/upper-constraints.txt#L475 | 16:39 |
openstackgerrit | John L. Villalovos proposed openstack/ironic-lib master: Use six.wraps() for Metrics so decorated methods can be inspected https://review.openstack.org/540031 | 16:42 |
sauloaislan | TheJulia: same error | 16:42 |
jlvillal | jroll, dtantsur rloo ^^^^ | 16:42 |
jlvillal | And I even did a full passing tox run at my end before I pushed. Amazing! :) | 16:43 |
dtantsur | oh, so six.wraps is different from functools.wraps? TIL | 16:44 |
dtantsur | jlvillal: could you please attach a bug to it? I guess we'll want a backport | 16:45 |
*** pcaruana has quit IRC | 16:45 | |
jlvillal | dtantsur, https://pythonhosted.org/six/ | 16:45 |
jlvillal | This is exactly the functools.wraps() decorator, but it sets the __wrapped__ attribute on what it decorates as functools.wraps() does on Python versions after 3.2. | 16:45 |
dtantsur | nice! | 16:46 |
jlvillal | dtantsur, Did someone already create a bug? | 16:46 |
dtantsur | not me for sure | 16:46 |
* jlvillal wanders over to launchpad | 16:46 | |
*** spotz has joined #openstack-ironic | 16:46 | |
jlvillal | https://bugs.launchpad.net/ironic/+bug/1746730 | 16:46 |
openstack | Launchpad bug 1746730 in Ironic "AgentRescue validation fails to get the signature of PXEBoot.prepare_ramdisk" [High,In progress] - Assigned to Jim Rollenhagen (jim-rollenhagen) | 16:46 |
*** bcornec has joined #openstack-ironic | 16:47 | |
openstackgerrit | John L. Villalovos proposed openstack/ironic-lib master: Use six.wraps() for Metrics so decorated methods can be inspected https://review.openstack.org/540031 | 16:47 |
jlvillal | dtantsur, ^^ done | 16:47 |
dtantsur | thnx | 16:48 |
rloo | dtantsur, jlvillal, jroll et al. the reason why the code in the conductor passed the unit tests is cuz the fake power did NOT have the metrics decorator :-( Otherwise, it would have failed. | 16:50 |
*** yee37933 has quit IRC | 16:51 | |
*** yee379 has joined #openstack-ironic | 16:51 | |
*** vabada has quit IRC | 16:52 | |
*** vabada has joined #openstack-ironic | 16:52 | |
openstackgerrit | Sam Betts proposed openstack/ironic master: Ensure we unbind flat network ports and clear mac addresses https://review.openstack.org/539981 | 16:52 |
rloo | even with the fix to ironic-lib, I am now a bit gun-shy about using reflection.get_signature() | 16:53 |
dtantsur | rloo: what if I remove the variant of FakePower without soft power? fixing it increases the patch quite substantially | 16:54 |
*** SerenaFeng has quit IRC | 16:55 | |
dtantsur | we have a lot of unit tests relying on fake_agent/fake_pxe using FakePower specifically | 16:55 |
TheJulia | sauloaislan: are you sure keystoneauth1 3.4.0 is no longer installed? | 16:55 |
rloo | dtantsur: i think we should decide how we want to handle this. we should definitely fix in ironic-lib. | 16:55 |
rloo | dtantsur: wrt ironic and use of reflection.... | 16:55 |
dtantsur | yep, this one I agree with | 16:55 |
dtantsur | now thinking about this story around fake power | 16:55 |
rloo | dtantsur: if we fix in ironic-lib, we can't use that lib version with master, right? | 16:55 |
rloo | dtantsur: cuz we can't release ironic-lib until next cycle | 16:56 |
dtantsur | rloo: yep, ironic master is now ironic-lib stable/queens | 16:56 |
dtantsur | meaning: we have to wait until ironic branches | 16:56 |
rloo | dtantsur: so do we do something temporary now, and undo in next cycle? | 16:56 |
rloo | dtantsur: or do we do something now that we don't undo? | 16:56 |
dtantsur | rloo: do we still need reflection around timeout? | 16:56 |
dtantsur | I think timeout was introduced long ago, we can just hardcode it | 16:57 |
*** rbartal has quit IRC | 16:57 | |
*** dsneddon has joined #openstack-ironic | 16:57 | |
rloo | dtantsur: that's a good question. i can take a look, i had filed bugs to get the intree drivers to support timeout. i think one of the cisco ones doesn't support it yet, something to do with needing to upgrade to a diff library, but i need to confirm. | 16:57 |
dtantsur | rloo: we can update cisco to accept the parameter and raise if it's not None | 16:58 |
* jroll doesn't think reflection is something that should be encouraged in production code, fwiw, except maybe down in the depths of some crazy metaprogramming | 16:59 | |
rloo | dtantsur: uhhh. i have 6 bugs open, none of them say that they have been released. | 16:59 |
dtantsur | sigh | 16:59 |
dtantsur | anyway, we can add no-op timeouts to all drivers, or we can make timeout!=None fail for them | 16:59 |
dtantsur | I don't think we should keep https://github.com/openstack/ironic/blob/master/ironic/conductor/utils.py#L229-L240 | 17:00 |
*** athomas has quit IRC | 17:00 | |
sambetts | what about out of tree drivers? I think that was the main reason for the inspection? | 17:00 |
sambetts | I don't know why we didn't fix the intree ones at the time of adding the paramter.. | 17:00 |
jroll | doesn't hurt to keep it for another cycle, it's been 3-4 cycles already. though use a try/except instead. | 17:01 |
*** athomas has joined #openstack-ironic | 17:01 | |
rloo | dtantsur: agree. we can't keep the reflection thing cuz it is broken. we can fix now i think by making call with the timeout ^^ as jroll mentions | 17:01 |
rloo | dtantsur: just wondering what we want to do going forward. | 17:01 |
dtantsur | I'd just go and break them.. if they haven't updated yet, they won't | 17:01 |
*** mjura has quit IRC | 17:02 | |
sambetts | i think there was a reason for using reflection instead of try/except I'm trying to remember why | 17:02 |
jroll | dtantsur: they didn't get the warning in the logs, though, was there a release note or something to indicate people should update? | 17:02 |
dtantsur | sambetts: because catching TypeError is too broad | 17:02 |
dtantsur | jroll: I don't remember, it was back in Ocata, I think | 17:02 |
dtantsur | I mean, you cannot do try: call() except TypeError: LOG.warning('plesae upgrade') | 17:03 |
dtantsur | you at least need to inspect the error message | 17:03 |
*** d0ugal has quit IRC | 17:03 | |
sambetts | yeah... :( | 17:04 |
jroll | dtantsur: not much of a release note here, I don't think operators would have noticed, honestly https://github.com/openstack/ironic/blob/master/releasenotes/notes/soft-reboot-poweroff-9fdb0a4306dd668d.yaml | 17:04 |
jroll | I wouldn't block just doing it but I support leaving it for a cycle since there wasn't very good notice | 17:04 |
jroll | er, I have it backwards. kill it with fire. | 17:05 |
jroll | we'd hit the case with the warning always. | 17:05 |
rloo | so if we just MakeTheCallWithTimeout, we could also add 'timeout' to the intree drivers with a warning there that it isn't supported. Would that work for everyone? | 17:07 |
openstackgerrit | Merged openstack/networking-generic-switch master: Updated from global requirements https://review.openstack.org/538502 | 17:07 |
rloo | fwiw, oneview, idrac, snmp, ucs, cimc, and ilo don't seem to support it (based on the bug reports) | 17:07 |
jroll | sigh | 17:08 |
rloo | and it breaks out of tree drivers. i wonder what is in staging-... | 17:08 |
dtantsur | we've approved oneview, but yeah.. | 17:08 |
*** harlowja has joined #openstack-ironic | 17:09 | |
*** devananda has joined #openstack-ironic | 17:11 | |
*** devananda has quit IRC | 17:12 | |
sauloaislan | TheJulia: Yes, in Ironic conductor, keystone, glance I change all to 3.1.0 | 17:12 |
*** devananda has joined #openstack-ironic | 17:12 | |
*** rloo has quit IRC | 17:15 | |
*** serlex has quit IRC | 17:15 | |
jlvillal | dtantsur, rloo: Do we want to backport the proposed ironic-lib fix to the stable branch? And then do a new release? | 17:16 |
jroll | jlvillal: we can't release queens libraries until after the ptg | 17:16 |
jroll | they're frozen until then | 17:16 |
jroll | or at least g-r updates are | 17:16 |
jlvillal | jroll, Ah | 17:16 |
jroll | might be possible to ffe it, not sure | 17:16 |
dtantsur | jroll: we can release queens for urgent fixes, but it won't affect master | 17:17 |
dtantsur | I think we cannot release master (which is rocky now for ironic-lib) | 17:18 |
jroll | ah, right right | 17:18 |
*** marios has quit IRC | 17:18 | |
*** rloo has joined #openstack-ironic | 17:18 | |
*** d0ugal has joined #openstack-ironic | 17:19 | |
dtantsur | so e.g. I plan on releasing sushy queens again, but that's because of external consumers | 17:19 |
rloo | dtantsur: are you just waiting on that one patch you backported? wrt sushy queens? | 17:24 |
dtantsur | yep | 17:24 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Migrate the remaining classic drivers to hardware types https://review.openstack.org/539589 | 17:24 |
dtantsur | rloo: I took an easy approach on fake stuff here ^^^ | 17:24 |
dtantsur | I really don't want to spend hours fixing tests :) | 17:24 |
rloo | dtantsur: heh | 17:26 |
rloo | dtantsur: btw, do we want to backport this: https://review.openstack.org/#/c/539582/ | 17:26 |
patchbot | patch 539582 - ironic - Add missing ilo vendor to the ilo hardware types (MERGED) | 17:26 |
TheJulia | rloo: to pike? | 17:27 |
dtantsur | rloo: it's on the border of being a feature | 17:27 |
dtantsur | I'd wait until somebody has a case for it | 17:27 |
TheJulia | +1 | 17:27 |
*** derekh has quit IRC | 17:27 | |
rloo | TheJulia, dtantsur: ok. it just seemed 'missing' to me, not a feature. but I'm good either way. | 17:27 |
dtantsur | rloo: yep, I'm not strictly -2, but I won't push for it | 17:28 |
rloo | dtantsur: got it. no one seems to have noticed that it was missing :) | 17:28 |
dtantsur | correct :) | 17:28 |
TheJulia | sauloaislan: I'm wondering how your getting that error, if configuration might be driving you to that error. Of course, I'm also confused by ironic version 17 in your paste... | 17:29 |
*** jtomasek_ has quit IRC | 17:31 | |
sauloaislan | The TheJulia: This paste is ironic master http://paste.openstack.org/show/653465/ | 17:32 |
sauloaislan | I have same error | 17:32 |
sauloaislan | keystoneauth1==3.4.0 | 17:32 |
dtantsur | jroll: wanna propose a PTG topic for something like "How detailed do we want to see our release notes?" | 17:34 |
jroll | dtantsur: idk, I hate bikeshedding things like that | 17:35 |
dtantsur | hehe | 17:35 |
jroll | don't feel like it's a productive use of expensive time | 17:35 |
rloo | ++ jroll, i'm curious as to what 'irrelevant' means. | 17:35 |
TheJulia | sauloaislan: ironic master should report 10.0.1.... not 17.0.0b3 | 17:35 |
rloo | jroll: so no need to mention that we changed the fake driver? | 17:35 |
rloo | jroll: and were there things you were hoping to see in the release notes, that weren't there? | 17:36 |
jroll | rloo: yeah, it just feels like sometimes our notes aren't very actionable, or necessary. but hard to say, everyone has a different perspective | 17:36 |
jroll | rloo: I don't think the release note about the fake driver will be useful to anyone. but hard to tell for sure. | 17:37 |
dtantsur | so, I'm working on a revision. should I delete the release note? | 17:37 |
jroll | rloo: and I won't know if there were missing things until I get further along in this upgrade :) | 17:37 |
jroll | dtantsur: +0.5 from me | 17:37 |
jroll | feels like noise | 17:37 |
rloo | dtantsur: no, i think we shoudl have the release note. i don't know if it is a feature though. maybe other? it is really minor. but it is a change. | 17:38 |
dtantsur | other works for me | 17:38 |
sauloaislan | TheJulia: My verion is ironic==9.2.1.dev89 | 17:38 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Migrate the remaining classic drivers to hardware types https://review.openstack.org/539589 | 17:39 |
dtantsur | rloo, jroll ^^^ | 17:39 |
jroll | +1 | 17:40 |
*** AlexeyAbashkin has quit IRC | 17:40 | |
jroll | left a nit as I couldn't resist | 17:40 |
jroll | (please don't fix it, zuul doesn't need the churn) | 17:40 |
dtantsur | lol | 17:40 |
*** AlexeyAbashkin has joined #openstack-ironic | 17:41 | |
* sambetts --> home | 17:41 | |
sambetts | night all | 17:41 |
*** sambetts is now known as sambett|afk | 17:41 | |
rloo | bye sambett|afk | 17:42 |
ricardoas | morning, ironic! are we frozen? | 17:42 |
ricardoas | :D | 17:42 |
* dtantsur looks outside | 17:42 | |
jroll | I think it's just above freezing where I am | 17:42 |
dtantsur | ricardoas: morning, yeah it's quite chilly here :) | 17:42 |
dtantsur | ricardoas: as to the feature freeze, we're finishing the last exceptions | 17:43 |
ricardoas | can´t say the same, jroll... :) | 17:43 |
dtantsur | mostly classic drivers deprecation and routed networks. see the whiteboard for details | 17:44 |
dtantsur | ricardoas: I guess it's quite warm where you are, yes :) and maybe even some sun | 17:44 |
ricardoas | great, dtantsur! That means no more time for bug fixes, right? | 17:44 |
dtantsur | ricardoas: actually, the very right time for bug fixes | 17:45 |
dtantsur | the freeze only affects features | 17:45 |
*** AlexeyAbashkin has quit IRC | 17:45 | |
*** trown is now known as trown|lunch | 17:46 | |
dtantsur | TheJulia: if you have a minute: https://review.openstack.org/#/c/537415/ | 17:47 |
patchbot | patch 537415 - ironic-inspector - Switch the CI to hardware types | 17:47 |
ricardoas | dtantsur: cool! :) | 17:48 |
openstackgerrit | Merged openstack/ironic master: Remove unused code in common/neutron.py https://review.openstack.org/538936 | 17:49 |
openstackgerrit | Merged openstack/ironic master: Follow-up for Switch OneView driver to hpOneView and ilorest libraries https://review.openstack.org/539237 | 17:49 |
openstackgerrit | Merged openstack/ironic master: Add documentation for baremetal mech https://review.openstack.org/537353 | 17:49 |
openstackgerrit | Merged openstack/ironic master: Update FAQ about updates of release notes https://review.openstack.org/539637 | 17:49 |
dtantsur | speaking of bugs, https://review.openstack.org/#/c/531180/ and https://review.openstack.org/#/c/531120/ could use (even non-core) reviews | 17:49 |
patchbot | patch 531180 - ironic - Handle case when a glance image contains no data | 17:49 |
patchbot | patch 531120 - ironic - Rework exception handling on deploy failures in co... | 17:49 |
TheJulia | dtantsur: done | 17:50 |
dtantsur | thnx | 17:52 |
dtantsur | side note: "no-rescue" sounds a bit hopeless :D | 17:58 |
*** amarao has quit IRC | 17:59 | |
*** amarao1 has quit IRC | 17:59 | |
TheJulia | but it is the truth? is it not? | 17:59 |
TheJulia | I mean, the only other thing we could do is have a doom interface which boots a freedos image and maybe fires up doom... but then we will need graphical console support | 18:00 |
* TheJulia is awkwardly surprised that this did not illicit a reply | 18:03 | |
jlvillal | rloo, TheJulia: Passed Zuul. https://review.openstack.org/540031 Issue with reflection.get_signature() | 18:05 |
patchbot | patch 540031 - ironic-lib - Use six.wraps() for Metrics so decorated methods c... | 18:05 |
rloo | thx jlvillal. although it isn't high on my list of priorities right now. i was just thinking i should see what else needs to land by tomorrow... | 18:06 |
jlvillal | Though unsure if priority or not. Should be fixed. | 18:06 |
jlvillal | rloo, Good plan. | 18:06 |
rloo | jlvillal: jroll's fix needs to land soon too. i think i already +2d it but i already forgot. | 18:06 |
*** mgoddard_ has quit IRC | 18:07 | |
dtantsur | TheJulia: Doom as a Service! open a start up! | 18:07 |
* dtantsur thinks how to apply blockchain here | 18:07 | |
*** baha has quit IRC | 18:08 | |
rloo | if anyone has a few minutes, this isn't hard to review and would be good to get merged soon: https://review.openstack.org/#/c/539961/ | 18:08 |
patchbot | patch 539961 - ironic - Remove mode argument from boot.(prepare|clean_up)_... | 18:08 |
hjensas | dtantsur: TheJulia: do we want platform specific stuff in docs? e.g ``systemctl enable ironic-neutron-agent`` ``systemctl start ironic-neutron-agent``? Or do we just say "ensure the service is running" and assume the reader knows how to do that on their distro? | 18:12 |
*** e0ne has joined #openstack-ironic | 18:14 | |
JayF | hjensas: At least when I was writing docs, we'd usually state the generic, then optionally include what you need to do; i.e. | 18:15 |
JayF | `ensure service X is running \n\n\t Ubuntu x.xx and newer: service blah start` or so on | 18:16 |
JayF | at least that's how we used to handle package installs | 18:16 |
rloo | TheJulia, dtantsur: so am I correct, there are no code patches that need to land by tomorrow wrt FFE, except dmitry's | 18:16 |
dtantsur | hjensas: ^^^? | 18:16 |
rloo | TheJulia, dtantsur: the rest look like docs that can land next week | 18:16 |
*** athomas has quit IRC | 18:16 | |
dtantsur | wow, a real JayF :) | 18:16 |
JayF | I'm always in IRC :) just never read more than a screenful of chat anymore when it gets my attention lol | 18:17 |
dtantsur | :) | 18:17 |
rloo | hiya JayF :) | 18:17 |
JayF | I'll hang around until all my features are merged | 18:17 |
dtantsur | hjensas: personally, I assume people know how to restart services, or they're unlikely to be able to install openstack manually... | 18:17 |
JayF | using my openstack timetable, that looks like 2038 or so /s | 18:17 |
hjensas | JayF: I see. I feel like leaving the details out, with all the containers and stuff its a long list of docker .. k8s, systemctl, service ... :) | 18:17 |
dtantsur | JayF: like, forever? | 18:17 |
JayF | dtantsur: you get the jokes :D | 18:17 |
dtantsur | hehe | 18:17 |
rloo | JayF: in that case, we'll make sure all but one lands :) | 18:17 |
dtantsur | rloo: we haven't even approached the specific faults work ;) | 18:18 |
rloo | dtantsur: oh yeah. so 2038 is about right :D | 18:18 |
*** openstackgerrit has quit IRC | 18:18 | |
*** baha has joined #openstack-ironic | 18:24 | |
*** rbudden has quit IRC | 18:24 | |
*** harlowja has quit IRC | 18:29 | |
*** rbudden has joined #openstack-ironic | 18:31 | |
*** vsaienk0 has quit IRC | 18:32 | |
*** openstackgerrit has joined #openstack-ironic | 18:37 | |
openstackgerrit | Merged openstack/python-ironicclient master: Zuul: Remove project name https://review.openstack.org/539741 | 18:37 |
TheJulia | hjensas: At this point, I think service restart is fine since we have no idea what the operating context is. | 18:38 |
*** dtantsur is now known as dtantsur|afk | 18:45 | |
dtantsur|afk | o/ | 18:45 |
TheJulia | goodnight dtantsur|afk | 18:46 |
openstackgerrit | Harald Jensås proposed openstack/ironic master: Fix configure-networking docs https://review.openstack.org/540085 | 18:47 |
*** e0ne has quit IRC | 18:49 | |
*** edleafe has quit IRC | 18:50 | |
*** rbudden has quit IRC | 18:50 | |
*** rbudden has joined #openstack-ironic | 18:58 | |
*** trown|lunch is now known as trown | 19:01 | |
*** edleafe has joined #openstack-ironic | 19:05 | |
*** edleafe has quit IRC | 19:07 | |
*** harlowja has joined #openstack-ironic | 19:09 | |
*** AlexeyAbashkin has joined #openstack-ironic | 19:11 | |
*** harlowja_ has joined #openstack-ironic | 19:11 | |
*** edleafe has joined #openstack-ironic | 19:12 | |
*** harlowja has quit IRC | 19:14 | |
*** vsaienk0 has joined #openstack-ironic | 19:15 | |
*** dprince has quit IRC | 19:19 | |
openstackgerrit | Olivier Bourdon proposed openstack/bifrost master: Add creation of openrc file when keystone enabled https://review.openstack.org/540096 | 19:20 |
openstackgerrit | Julia Kreger proposed openstack/ironic master: Don't try to lock for vif actions https://review.openstack.org/534441 | 19:31 |
openstackgerrit | Julia Kreger proposed openstack/networking-baremetal master: Add dsvm job https://review.openstack.org/530409 | 19:34 |
*** vsaienk0 has quit IRC | 19:39 | |
*** sauloaislan has quit IRC | 19:42 | |
*** vsaienk0 has joined #openstack-ironic | 19:50 | |
*** lucasagomes is now known as lucas-afk | 19:55 | |
*** AlexeyAbashkin has quit IRC | 19:55 | |
openstackgerrit | Merged openstack/ironic-inspector master: Switch the CI to hardware types https://review.openstack.org/537415 | 19:56 |
*** dyasny has left #openstack-ironic | 19:56 | |
*** vsaienk0 has quit IRC | 20:00 | |
TheJulia | I have updated the whiteboard to represent current statuses | 20:03 |
TheJulia | I think the networking-braemetal stuff is the only outstanding code aside from bugs we've spotted | 20:04 |
*** dprince has joined #openstack-ironic | 20:04 | |
openstackgerrit | Merged openstack/ironic master: Devstack - use neutron segments (routed provider networks) https://review.openstack.org/531275 | 20:06 |
*** d0ugal has quit IRC | 20:13 | |
ricardoas | cd | 20:18 |
ricardoas | oops... sorry | 20:18 |
TheJulia | ricardoas: no worries! :) | 20:23 |
*** d0ugal has joined #openstack-ironic | 20:24 | |
rloo | thx TheJulia. This seems to be the only/main outstanding one for routed networks, but I don't feel qualified to review w/o spending time getting up to speed: https://review.openstack.org/#/c/524709/ | 20:33 |
patchbot | patch 524709 - networking-baremetal - Make the agent distributed using hashring and noti... | 20:33 |
TheJulia | yeah, hash ring stuff makes my head hurt.... the DSVM job should land before we even try to merge that though | 20:35 |
jroll | I can make an attempt at that one | 20:36 |
rloo | TheJulia: this dsvm job? https://review.openstack.org/#/c/530409/ | 20:37 |
patchbot | patch 530409 - networking-baremetal - Add dsvm job | 20:37 |
openstackgerrit | Merged openstack/bifrost master: Rewrite iptables rules with Ansible module https://review.openstack.org/536911 | 20:37 |
openstackgerrit | Merged openstack/ironic master: Do not use asserts with business logic https://review.openstack.org/538991 | 20:38 |
openstackgerrit | Merged openstack/ironic master: Add api-ref and ironic state documentation for rescue https://review.openstack.org/419606 | 20:38 |
rloo | but that depends on this: https://review.openstack.org/#/c/531637/ | 20:38 |
patchbot | patch 531637 - ironic - Wait for ironic-neutron-agent to report state | 20:38 |
* rloo runs away | 20:38 | |
rloo | thx jroll :) | 20:38 |
jroll | I like how it's using oslo.messaging to track membership, that's interesting :) | 20:40 |
hjensas | jroll: I'll take that as a compliment. :D | 20:41 |
*** ricardoas is now known as ricardoas-away | 20:49 | |
*** baoli has joined #openstack-ironic | 20:50 | |
*** baoli has quit IRC | 20:52 | |
openstackgerrit | Merged openstack/ironic master: Remove mode argument from boot.(prepare|clean_up)_ramdisk https://review.openstack.org/539961 | 20:56 |
jroll | hjensas: :) left a few comments, but it's close | 20:58 |
jroll | and a couple more | 21:00 |
hjensas | jroll: thanks, yes I was looking at BackoffLoopingCall before. I will look into it later. | 21:01 |
jroll | hjensas: cool | 21:02 |
*** priteau has quit IRC | 21:03 | |
*** ianychoi_ has joined #openstack-ironic | 21:21 | |
openstackgerrit | Olivier Bourdon proposed openstack/bifrost master: Add creation of openrc file when keystone enabled https://review.openstack.org/540096 | 21:22 |
*** dprince has quit IRC | 21:22 | |
*** ianychoi has quit IRC | 21:24 | |
*** Goneri has quit IRC | 21:32 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic master: Remove sample policy and config files https://review.openstack.org/536349 | 21:33 |
jroll | ^ we should land that before we keep conflicting with it, dtantsur|afk was already +2 | 21:33 |
rloo | jroll: i was waiting till all the FFE patches landed that might conflict with that. i think we're good now. | 21:38 |
jroll | rloo: yeah, makes sense, I checked that before asking too :) | 21:39 |
rloo | jroll: commented | 21:45 |
*** olivierbourdon has joined #openstack-ironic | 21:45 | |
jroll | rloo: I'll add that, why not | 21:49 |
openstackgerrit | Ruby Loo proposed openstack/ironic master: Fix handling of optional 'timeout' parameter to power methods https://review.openstack.org/540150 | 21:50 |
rloo | thx jroll | 21:50 |
rloo | and that ^^ was a pain.in... and I'm not finished with it :-( | 21:50 |
*** olivierbourdon38 has joined #openstack-ironic | 21:50 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic master: Remove sample policy and config files https://review.openstack.org/536349 | 21:54 |
jroll | ouch, looks painful | 21:54 |
*** olivierbourdon has quit IRC | 21:55 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost master: Deprecate Classic Drivers https://review.openstack.org/540153 | 21:56 |
*** threestrands has joined #openstack-ironic | 21:59 | |
*** threestrands has quit IRC | 22:00 | |
*** threestrands has joined #openstack-ironic | 22:00 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost master: Fedora 25-27 support https://review.openstack.org/471750 | 22:00 |
*** trown is now known as trown|outtypewww | 22:01 | |
olivierbourdon38 | Any known issue with bifrost deployment on Ubuntu 16.04 with keystone enabled ? Seems like commands provided by python-openstackclient are missing in case of no VENV used and keystone-manage missing when using VENV ? | 22:03 |
olivierbourdon38 | I managed to fix the 1st one by installing the appropriate git repo after keystone one but wanted to know about this before going on fixing the second one | 22:03 |
openstackgerrit | Merged openstack/ironic master: Migrate the remaining classic drivers to hardware types https://review.openstack.org/539589 | 22:05 |
jroll | \o/ | 22:06 |
jroll | have a good night y'all | 22:06 |
openstackgerrit | Julia Kreger proposed openstack/bifrost master: Fedora 25-27 support https://review.openstack.org/471750 | 22:10 |
TheJulia | goodnight jroll | 22:10 |
TheJulia | olivierbourdon38: heh, I guess your in a bit of a case that we didn't test for :\ | 22:11 |
olivierbourdon38 | no pb, I can submit the patch after loggin a launchpad bug if you want | 22:12 |
TheJulia | olivierbourdon38: but openstack command should be present on ubuntu 16.04 w/o a VENV.... Odd | 22:12 |
TheJulia | olivierbourdon38: by all means! It is greatly appreciated! | 22:12 |
olivierbourdon38 | may be the reason is that I am using a very minimal (JEOS) ubuntu image based on OpenSSH server + few others and I was relying on env-setup and ansible to do the rest | 22:14 |
TheJulia | That might explain some of it.... | 22:14 |
TheJulia | olivierbourdon38: feel free to ping me with the bug and patch links, I'll be around for a couple more hours and can take a look. I think 16.04 is what I have running on one of my machines | 22:15 |
olivierbourdon38 | i need a few more time testing my patches and as it is already quite late in my TZ I'll leave that for tomorrow morning ;-) many thx | 22:16 |
TheJulia | olivierbourdon38: okay, have a wonderful evening! :) | 22:20 |
*** baha has quit IRC | 22:23 | |
jlvillal | rloo, I just reviewed https://review.openstack.org/#/c/539676/ Ping me if I should remove the -1 | 22:35 |
patchbot | patch 539676 - ironic - [api-ref] clarify what /v1/lookup returns | 22:35 |
rloo | jlvillal: to be honest, i have no idea and don't intend on looking any further. i was just clarifying hwat the subset of fields were. | 22:36 |
jlvillal | rloo, Okay. Well in the spirit of "is it making the code better?". I think it is. | 22:38 |
rloo | jlvillal: thx, I replied. I think if we wanted to mention links, someone should do it throughout so it is consistent. i think we always return links with nodes. | 22:40 |
jlvillal | rloo, Thanks! | 22:41 |
openstackgerrit | Merged openstack/ironic master: Updated from global requirements https://review.openstack.org/538463 | 22:43 |
openstackgerrit | Merged openstack/ironic-python-agent master: Adds a releasenote for DHCP support for TinyIPA https://review.openstack.org/539639 | 22:44 |
*** bfournie has quit IRC | 22:46 | |
*** olivierbourdon38 has quit IRC | 22:54 | |
*** jcoufal_ has quit IRC | 22:58 | |
*** olivierbourdon38 has joined #openstack-ironic | 23:02 | |
*** priteau has joined #openstack-ironic | 23:03 | |
*** cdearborn has quit IRC | 23:05 | |
*** priteau_ has joined #openstack-ironic | 23:07 | |
*** priteau has quit IRC | 23:08 | |
*** priteau_ has quit IRC | 23:11 | |
openstackgerrit | John L. Villalovos proposed openstack/ironic-python-agent master: Replace use of functools.wraps() with six.wraps() https://review.openstack.org/540175 | 23:18 |
*** bfournie has joined #openstack-ironic | 23:30 | |
hjensas | TheJulia: I replied to comment from vsaienko on https://review.openstack.org/#/c/531637 - not much we can do. | 23:34 |
patchbot | patch 531637 - ironic - Wait for ironic-neutron-agent to report state | 23:34 |
TheJulia | I had a feeling that would be the case | 23:34 |
TheJulia | thanks! you rock! | 23:34 |
*** r-daneel has quit IRC | 23:35 | |
hjensas | You too! :) | 23:35 |
*** MattMan has quit IRC | 23:36 | |
*** MattMan has joined #openstack-ironic | 23:36 | |
*** hshiina|afk has joined #openstack-ironic | 23:42 | |
*** rbudden has quit IRC | 23:43 | |
*** charcol-v2 has joined #openstack-ironic | 23:56 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!