[Openstack-operators] need input on log translations

Florian Haas florian at hastexo.com
Sun Mar 12 19:01:57 UTC 2017


On Fri, Mar 10, 2017 at 6:19 PM, Amrith Kumar <amrith.kumar at gmail.com> wrote:
> Today the workflow is to copy the message and plonk it into google and let
> search do its thing.
>
> Given this, log message translation fragments the solution space because
> nothing out there consolidates the reports of problems with error messages
> and converts the non-english messages back to some common language for
> search.
>
> I'm not an operator, but I'd love to hear from operators whether this
> workflow is in fact the one we should be planning for because it has
> implications that go beyond translation. First, should we make messages
> uniquely searchable, such that a search result set uniquely relates to a
> specific problem. Conversely, if we go the route of message translation in
> the future should all messages include some searchable code (like Microsoft
> does with their crash codes).

Native speaker of German here, and not an operator either but someone
who frequently consults with operators including rather advanced
troubleshooting engagements. I agree entirely with what Amrith is
saying here. Solution space fragmentation is bad; log translation
contributes to that fragmentation. And yes having dependable unique
identifiers on error/warning/notice conditions would be oh so awesome.

Cheers,
Florian



More information about the OpenStack-operators mailing list