Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-04-08
02:57:41 ileixe How can admin achieve their own busineess logic with VM lifecycle? For example, when VM created, add related metadata to CMDB.
02:58:11 ileixe We've been used notification events before, I realized it's quite unstable and unexpectable..
04:15:43 openstackgerrit Sundar Nadathur proposed openstack/nova-specs master: Nova Cyborg interaction specification. https://review.openstack.org/603955
05:50:54 openstackgerrit Boxiang Zhu proposed openstack/nova master: Make evacuation respects anti-affinity rule https://review.openstack.org/649953
06:36:26 openstackgerrit Boxiang Zhu proposed openstack/nova master: Use the functional test test_parallel_evacuate_with_server_group https://review.openstack.org/649963
06:38:29 openstackgerrit zhufl proposed openstack/nova master: Remove query_client from resource_tracker https://review.openstack.org/650616
07:03:46 openstackgerrit Kunpeng Zhang proposed openstack/nova master: Enhance live-migration progress log for memory https://review.openstack.org/650621
07:09:21 openstackgerrit Tushar Patil proposed openstack/nova-specs master: Allow compute nodes to use DISK_GB from shared storage RP https://review.openstack.org/650188
07:13:47 openstackgerrit Kunpeng Zhang proposed openstack/nova master: Enhance live-migration progress log for memory https://review.openstack.org/650621
07:14:44 openstackgerrit Takashi NATSUME proposed openstack/nova stable/rocky: doc: Fix openstack CLI command https://review.openstack.org/648425
07:15:04 openstackgerrit Takashi NATSUME proposed openstack/nova stable/rocky: Replace glance command with openstack command https://review.openstack.org/637060
07:28:33 gibi_off ileixe: have you reported the problems regarding notifications as bugs? I encurage you to try versioned notifications and report any bugs. Also feel free to ping me if you have questions
07:31:34 openstackgerrit Yongli He proposed openstack/nova master: Clean up orphan instances https://review.openstack.org/627765
07:31:52 openstackgerrit Takashi NATSUME proposed openstack/nova master: Change a log level for overwriting allocation https://review.openstack.org/649788
07:34:50 ileixe gibi: I think I do not choose right word to express my situation. Notification itself is not a problem. What I felt hard is whenver I have small hooking logic, there were too many things to be considered. (e.g. We have to deploy ceilometer for consumer, gnocchi for publisher, aodh for alarm. And even there is no exact API to hook the event)
07:35:35 ileixe I felt like I'm on the wrong track. T_T
07:37:04 gibi ileixe: have you considered simply listening on the notification (or versioned_notification) topic on the message bus? that needs credentials to the message bus but if you can have that then it is couple of lines of code to subscribe to the topic and get the events
07:37:56 ileixe Yes acutally we're done to make our 3rd party component to listen messaging
07:39:01 gibi in this case you dont need ceilometer and aodh to interface with your CMDB
07:39:25 ileixe but I thought it's also not a right solution since we have to track messaging format/every event type (I mean unexpectable in this siutation).
07:39:35 ileixe Yep..
07:40:08 ileixe So my question is how others deal with this problem.
07:42:17 gibi ileixe: versioned notifications have a well defined format and proper versioning. Have you see the doc https://docs.openstack.org/nova/latest/reference/notifications.html#existing-versioned-notifications ?
07:43:21 gibi I cannot say how others deal with this, but I think versioned notificatons are capable of solving your problem and if you have feedback how to make them better then that feedback is very much appreciated
07:49:01 ileixe gibi: Thanks, gibi. It's worth to listen your view since I felt maybe I'm not going the wrong way. I will feedback you when I have an better idea to solve the problem.
07:49:15 gibi ileixe: feel free to ping me
07:49:25 ileixe :)
08:17:35 openstackgerrit Takashi NATSUME proposed openstack/nova master: Adds view builders for keypairs controller https://review.openstack.org/347289
08:17:58 openstackgerrit Takashi NATSUME proposed openstack/nova master: Fix 500 error while passing 4-byte unicode data https://review.openstack.org/407514
08:25:26 openstackgerrit Boxiang Zhu proposed openstack/nova master: Make evacuation respects anti-affinity rule https://review.openstack.org/649953
08:33:03 openstackgerrit Boxiang Zhu proposed openstack/nova master: Use the functional test test_parallel_evacuate_with_server_group https://review.openstack.org/649963
09:11:48 openstackgerrit Adrian Chiris proposed openstack/nova master: SR-IOV Live migration indirect port support https://review.openstack.org/620115
09:36:13 openstackgerrit Merged openstack/nova master: Do not log a warning about not using compute monitors https://review.openstack.org/650092
09:42:28 kashyap stephenfin: When you get a moment, an rST question: in this rendering: http://logs.openstack.org/14/645814/2/check/openstack-tox-docs/0e76299/html/specs/train/approved/cpu-selection-with-hypervisor-consideration.html
09:43:21 kashyap stephenfin: ... I don't like how the words "or" and "type" are highlighted in the two tables. (Probably because they're rendered as code blocks.)
09:43:58 kashyap Any ideas to address that?
09:45:32 openstackgerrit Kashyap Chamarthy proposed openstack/nova-specs master: Add "CPU selection with hypervisor consideration" spec https://review.openstack.org/645814
10:00:24 stephenfin It's trying to guess the syntax. You can do '.. code-block:: none' before it (instead of '::') to override that, I would say
10:02:14 kashyap Oh right!
10:02:19 kashyap stephenfin: I recall the 'none' thingie. Good idea.
10:02:28 kashyap Thanks :-)
10:03:16 stephenfin kashyap: Actually, looks like Sphinx can render complex tables like those. Left comment inline
10:04:04 kashyap Oh, yeah. I recalled the "+" syntax. Wonderful rST + Sphinx. Endless joy :-)
10:12:00 kashyap stephenfin: Okay, even with '+', the words "or" and "type" are highlighted in the rendering.
10:12:31 kashyap (NB: Rendering was not the problem; with +, it rendering was equivalent to the one before.)
10:13:08 stephenfin kashyap: If you switch that to '+', you can dedent it and remove the '::' or '.. code-block'
10:13:19 stephenfin so it'll render as a table instead of a code block
10:14:37 kashyap Ah-ha, so after switching it required 'dedenting'...
10:14:50 stephenfin Yeah, sorry, I wasn't clear
10:15:23 stephenfin The reason it's indented is so it'll render as code because I didn't think docutils would handle complex tables like that, but it turns out it can
10:15:27 stephenfin and with gusto, no less
10:15:35 kashyap (No problem.)
10:15:37 kashyap Hehe, noted.
10:17:38 openstackgerrit Bhagyashri Shewale proposed openstack/nova master: Ignore root_gb for BFV in simple tenant usage API https://review.openstack.org/612626
10:22:16 cdent "with gusto"
10:24:38 kashyap stephenfin: Hmm, so with "+" (de-dendeted, and no '::' or '.. code-block'), the table rendering is "naked": https://kashyapc.fedorapeople.org/Plain-rendering-of-table-Sphinx.png
10:24:49 kashyap (Assuming if you can even spot the table :D)
10:26:55 sean-k-mooney o/
10:31:44 sean-k-mooney stephenfin: efried: gibi: jaypipes: ill send a email about this later but https://review.openstack.org/#/c/650573/ as an fyi i finished setting up my thirdparty ci over the weekend so we have limited capastity to trigger some dpdk and pinning testing manually now.
10:33:17 sean-k-mooney if people comment "seans-nfv-ci: recheck" or "check experimental" on any patch to nova,neutron,os-vif,networking-ovs-dpdk or devstack then it will run
10:34:15 stephenfin sounds good to me
10:34:33 sean-k-mooney i currently have two singel node jobs running for nova one that tests ovs-dpdk(with hugepages) and another that tests numa+cpu pinning+realtime
10:36:19 sean-k-mooney im going to convert them to multinode jobs so that migration/resize tests run and add a whitbox plugin job too i think.
10:38:17 sean-k-mooney i also need to enable more tests. ill proably change the trigger to run atomatically once i come up with a sutible file ignore list but i am going to see if i can run some of these test upstream too in the future.
10:55:42 efried sean-k-mooney: nice!
11:03:52 openstackgerrit Eric Fried proposed openstack/nova-specs master: Re-propose detach root volume spec https://review.openstack.org/650580
11:08:24 openstackgerrit Eric Fried proposed openstack/nova-specs master: Re-propose volume backed server rebuild https://review.openstack.org/650582
11:19:27 openstackgerrit Eric Fried proposed openstack/nova master: Hacking N363: Don't use spec[_set]='string' https://review.openstack.org/650370
11:20:49 openstackgerrit Merged openstack/nova-specs master: Re-propose detach root volume spec https://review.openstack.org/650580
11:24:58 openstackgerrit Merged openstack/nova-specs master: Re-propose volume backed server rebuild https://review.openstack.org/650582
11:38:53 kashyap sean-k-mooney: sfinucan: Remind me again: why are we messing with VNC whe configuring live migration here: http://git.openstack.org/cgit/openstack/nova/tree/doc/source/admin/configuring-migrations.rst
11:40:25 sean-k-mooney kashyap by default our install guides did not used to tell people to make vnc listen to 0.0.0.0 i think
11:40:45 sean-k-mooney as a result if you live migrated without haveing mand that change the console would not work on the new host
11:41:12 sean-k-mooney as it defaulted to listening to on the ip of the source host and we dont update that in the libvirt xml on migration
11:41:42 sean-k-mooney so the vnc consle woudl try to bind to a tcp port on a ip that is not on the dstionation and fail
11:43:15 kashyap Ah, I see. I think we should spell that reasoning out in the doc. (I'll send a patch with "Thanks: Sean Mooney")
11:43:54 sean-k-mooney well it says "You must not make the VNC server listen to the IP address of its compute
11:43:56 sean-k-mooney host, since that addresses changes when the instance is migrated.
11:43:58 sean-k-mooney "
11:44:19 sean-k-mooney but ya it could be clearer
11:45:24 lyarwood sean-k-mooney / kashyap ; isn't that out of date now given https://github.com/openstack/nova/blob/65874326404717b58ce4927c21d3502d7e53022b/nova/virt/libvirt/migration.py#L84
11:45:41 kashyap sean-k-mooney: Yeah, the meaning is implicit. But it can definitely be clearer
11:45:59 lyarwood I thought we did update the listen address now
11:46:21 sean-k-mooney we do according to that code
11:46:35 sean-k-mooney so its nolonger requried but it might be desirable anyway
11:46:44 sean-k-mooney it may not be
11:46:52 sean-k-mooney when was that added
11:47:15 sean-k-mooney 3 years ago
11:47:36 sean-k-mooney there is a slight change that the dock is out dated
11:48:07 kashyap Yeah, definitely making sure that is set does not hurt.
11:48:18 kashyap lyarwood: Thanks for the pointer
11:56:50 kashyap sean-k-mooney: Do we always go through VNC proxy?
11:57:34 sean-k-mooney technically you can connect directly but typically an enduser will always connect via the vnc proxy
11:57:52 sean-k-mooney we do not want endusers connecting directly to the hyperviors
11:57:55 sean-k-mooney host
12:02:34 kashyap Nod; also how about security? Isn't _always_ listening on 0.0.0.0 insecure?
12:03:04 sean-k-mooney yes which is why the note is present in the doc.
12:03:18 kashyap Which note? /me reads
12:03:29 kashyap Ah, this one:
12:03:30 kashyap Since this setting allows VNC clients from any IP address to connect to instance consoles, you must take additional measures like secure networks or firewalls to prevent potential attackers from gaining access to instances.

Earlier   Later