Sean Mooney smooney at
Tue Aug 3 14:05:22 UTC 2021

On Tue, 2021-08-03 at 12:29 +0000, Jeremy Stanley wrote:
> On 2021-08-03 13:02:45 +0100 (+0100), Sean Mooney wrote:
> > it might make more sense to eventually move this to somethign
> > hosted by opendev rather then relying on the good nature of the
> > comunity to run it.
> [...]
> We did try at one time some years ago, but the service was
> unmanageable (required a restart to update any of the user data,
> which took hours and up to a day to re-query everything and come
> back online) and buggy (left hung SSH connections in its wake until
> the max sessions limit enforced for the Gerrit API kicked in and
> blocked it from reconnecting). We had volunteers lined up to solve
> these problems, and even some patches already proposed to start
> addressing them, but could not get the maintainers to review that
> work. Eventually we gave up, scrapped the idea, and tore down our
> server for it.
> Unless those issues have been addressed, we'd need new volunteers to
> get the service in shape... but even so, I'm hesitant to suggest
> OpenDev take on maintaining new services when we're actively trying
> to shed responsibilities so we can focus on improving services the
> community relies strongly on to get things done.

ack capasity is a concern but coudl we host a simple redirect or cname so that
if the serivce moves again we dont need to cahnge our messaging to contibutors.
cnames would have ssl issues but a rediect would work to atleast land people on the correct verions
that currently works. perhaps that is too much work for little benifit.

honestly im not sure how much upstram porject actully use it anymore. i know that
some compainies do use the data from stackalitics for incentives but honestly i think
that actully harms the comunity as it can promote trivial or nuisance commits jus tto games stats.
by nuisance commit i dont actully mean commits that dont fix anything i mean when the same typo
fix is summit to evey repo or where commit are broken down in many one of small line count commits to fix effectivly
the same thing that could have been grouped into one commit.

there are some good uses of stakalitics like evaluating the healt of a project by its vilocity, diverity and such
so i think its still valuable. but it can be abused too.

More information about the openstack-discuss mailing list