| *** sfinucan is now known as stephenfin | 09:01 | |
| amorin | hey nova team, is there any way to not rely on cell_mappings (from db) but only read data from the config file? | 09:02 |
|---|---|---|
| *** elodilles is now known as elodilles_afk | 10:00 | |
| sean-k-mooney | amorin: no | 10:58 |
| sean-k-mooney | there is a feature called templated cell db urls | 10:59 |
| sean-k-mooney | that allwos the username and password for the cell to be stored in the config but that woudl require using the same username/password for all cells | 10:59 |
| sean-k-mooney | amorin: https://docs.openstack.org/nova/latest/admin/cells.html#template-urls-in-cell-mappings | 11:00 |
| jlejeune | hello, it seems that since the last patchs related to the bug #2059809, you have added a new dependency with qemu-utils package to be able tu run qemu-img binary, I have two failling tests related to that: nova.tests.unit.image.test_format_inspector.TestFormatInspectors.test_vmdk_bad_descriptor_mem_limit and | 11:28 |
| jlejeune | nova.tests.unit.image.test_format_inspector.TestFormatInspectors.test_vmdk_bad_descriptor_mem_limit_stream_optimized. Are you aware of that ? Maybe a missing dependency in test-requirements.txt ? | 11:28 |
| jlejeune | or maybe these tests are not correctly skipped if qemu-img is not present in PATH | 11:34 |
| jlejeune | for other tests | 11:34 |
| opendevreview | sean mooney proposed openstack/nova master: [DNM] Test bumping guest ram https://review.opendev.org/c/openstack/nova/+/924142 | 11:38 |
| sean-k-mooney | jlejeune: this was a preexisting depency | 11:43 |
| sean-k-mooney | oh in the unit test | 11:43 |
| sean-k-mooney | am we added some skip patches | 11:43 |
| sean-k-mooney | to skip if qemu was not there | 11:44 |
| sean-k-mooney | its posible that we have some paths that are still not skipped | 11:44 |
| jlejeune | indeed, in self._create_allocated_vmdk() method in unit test | 11:44 |
| jlejeune | patching if you want | 11:45 |
| sean-k-mooney | you can submit a patch for that if you have time | 11:45 |
| jlejeune | sure, doing | 11:45 |
| sean-k-mooney | its not strictly speaking incorrect | 11:45 |
| sean-k-mooney | it just nova tened to try and have depencies on external tools in our unit tests | 11:46 |
| sean-k-mooney | glance however does not have that concept | 11:46 |
| sean-k-mooney | so when we import the unit tests form glance we gained this depency and we have been trying to relax it to make it optional | 11:46 |
| sean-k-mooney | eventurlaly these tests will be remove form nova and moved to oslo when the image inspector is moved | 11:46 |
| jlejeune | ok get it | 11:47 |
| amorin | sean-k-mooney, that's perfect, it will fit my needs | 11:54 |
| *** elodilles_afk is now known as elodilles | 12:07 | |
| *** haleyb|out is now known as haleyb | 12:32 | |
| opendevreview | Julien LE JEUNE proposed openstack/nova master: Fix test_vmdk_bad_descriptor_mem_limit and test_vmdk_bad_descriptor_mem_limit_stream_optimized https://review.opendev.org/c/openstack/nova/+/924148 | 12:47 |
| amorin | hey again, another question regarding BDM, when booting an instance with a BDM, I cannot find a way to specify the availability-zone field, while if I create the volume before using cinder API, it's available, is it missing on nova side on purpose? | 13:42 |
| ralonsoh | sean-k-mooney, hello! If you have some mins, please check https://review.opendev.org/c/openstack/neutron/+/923962 | 14:53 |
| ralonsoh | is that something you (nova team) discussed with the author? | 14:54 |
| opendevreview | Pranali Deore proposed openstack/nova master: DNM: Test glance new location api https://review.opendev.org/c/openstack/nova/+/891207 | 14:56 |
| sean-k-mooney | so OVS_HYBRID_PLUG is not intented to be used with ovn | 15:04 |
| sean-k-mooney | ralonsoh: i have not read the patch properly but i wound not normally expect that to be set when using ovn | 15:05 |
| sean-k-mooney | we wait for port binind events dirfferencly for ovn | 15:06 |
| opendevreview | Julien Le Jeune proposed openstack/nova master: Fix test_vmdk_bad_descriptor_mem_limit and test_vmdk_bad_descriptor_mem_limit_stream_optimized https://review.opendev.org/c/openstack/nova/+/924148 | 15:32 |
| sean-k-mooney | amorin: yes its not in the api bdm because if the volume is created by nova it will have the same az as the instance and if its a prexisting volume form cinder then it will already be recored on the volume | 17:59 |
| chris218 | Hi, does placement track usage of vcpu/pcpu cores on given node or it's done by nova-compute? I was wondering if it would be possible to run two nova-compute instances on one node with different drivers configured on them | 20:31 |
| noonedeadpunk | nova-compute does report consumption data to the placement iirc | 20:33 |
| noonedeadpunk | well, the tricky thing might be to make nova-compute service to register independently | 20:34 |
| noonedeadpunk | as it relies on hostnames. while in nova.conf you can set different `my_ip` and `host` vars per service, I'm not 100% sure you will be able to do the same trick for hypervisors... As there table is populated by what driver would report as a hostname... | 20:36 |
| noonedeadpunk | and then this will be pushed towards placement | 20:37 |
| chris218 | so... yes but it's complicated | 21:07 |
| opendevreview | Rajat Dhasmana proposed openstack/nova master: Rebuild server: Add offset for iSCSI volumes https://review.opendev.org/c/openstack/nova/+/924184 | 21:47 |
| melwitt | amorin: if nova is creating the volume, it will create it in the same AZ as you specify you want to create the instance. it's mentioned here (and maybe also in other docs) https://docs.openstack.org/nova/latest/configuration/config.html#cinder.cross_az_attach | 21:53 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!