On 07/27/2018 10:07 AM, Dean Troyer wrote:
On Fri, Jul 27, 2018 at 10:54 AM, Jones, Bruce E <bruce.e.jones@intel.com <mailto:bruce.e.jones@intel.com>> wrote:
In the short term, if there are logs that should be captured, please list them in the template and let’s figure out how to post them somewhere – we can’t post files to Stories or Tasks Lbut we can post links to them.
This is usually a pastebin, we have paste.openstack.org <http://paste.openstack.org> available...
Customers often don't know what logs are important and don't necessarily have the expertise to extract key snippets. Accordingly, our "collect" script basically grabs much of /etc/ and /var/log from each specified node, as well as database dumps on the controller node (and other information as well). This can easily add up to hundreds of MB of compressed tarball. We currently stick these on a shared internal filesystem and mention the location in the internal bug tracker. I wonder if the "collect" tool should have the option to upload logs to a "support" server and return an identifier to the customer to include in their bug report. If you just allow anyone to dump large files on a server somewhere it risks being turned into a filesharing site. The Cinder and Manila folks have also highlighted the lack of support for attachments as a serious problem: https://storyboard.openstack.org/#!/story/2003071 Chris