On 12/12/19 6:09 PM, Albert Braden wrote:
This is the Rocky doc I was using:
https://docs.openstack.org/neutron/rocky/configuration/neutron.html
This is what it says about the rpc_workers setting:
rpc_workers
Type: integer Default: 1 Number of RPC worker processes for service.
Just an FYI that the default value for rpc_workers has changed, if you look at the Stein or later config it's now: api_workers Type integer Default <None> Number of separate API worker processes for service. If not specified, the default is equal to the number of CPUs available for best performance, capped by potential RAM usage. rpc_workers Type integer Default <None> Number of RPC worker processes for service. If not specified, the default is equal to half the number of API workers.
Is this the correct place to add something like 'If your neutron-metadata-agent.log is full of "error: [Errno 32] Broken pipe" then increase this setting' ?
I'd be happy to help update the docs but I'm not sure how to get started. Opening document bugs doesn't seem to result in any change; if I could figure out how to submit changes myself that might be more helpful. How can I register as a document change submitter?
There is a doc in the neutron repo available via docs.o.o: https://docs.openstack.org/neutron/train/admin/config-wsgi.html The last section of that covers "Neutron Worker Processes", and might be helpful. If you think it would be useful to add a little more info on this failure please either create a patch (file is at doc/source/admin/config-wsgi.rst in the repo), or open a bug - sometimes these low-hanging fruit things are great for someone new to the project. -Brian
-----Original Message----- From: Matt Riedemann <mriedemos@gmail.com> Sent: Thursday, December 12, 2019 2:38 PM To: openstack-discuss@lists.openstack.org Subject: Re: Finding configuration values
On 12/12/2019 3:34 PM, Albert Braden wrote:
This illustrates a problem I've seen in Openstack. Many of the configuration settings are undocumented, or the documentation is obscure. The only way for an openstack operator to know the correct setting is to be told by an openstack developer.
Since Pike the docs in each project should be standardized and the configuration options should be generated from code. So for example: