I've noticed in my time on this list that it seems like the reliability of Wikimedia Analytics services is a bit spotty. I've worked in IT services, and our web and email servers' reliability seemed pretty good, comparable to the Wikimedia content delivery services. I'm curious if there is something about the nature of Analytics services that makes them inherently fragile, or if there is something about Wikimedia's particular configuration that is an issue. This isn't intended as criticism; I'm just curious.
Thanks for your work keeping this place running.
Pine
On Oct 12, 2014 8:49 PM, "Pine W" wiki.pine@gmail.com wrote:
I've noticed in my time on this list that it seems like the reliability
of Wikimedia Analytics services is a bit spotty. I've worked in IT services, and our web and email servers' reliability seemed pretty good, comparable to the Wikimedia content delivery services.
Who is "our"?
Which services do you have in mind?
Some services (e.g. wikimetrics, reportcard) run on labs so they inherent the lower uptime expectations of labs vs. prod.
Some services were historically not very reliable but should soon be more reliable than they ever were before. (see footnote 1 at bottom of https://lists.wikimedia.org/pipermail/analytics/2014-October/002539.html ; subject=Adventures in Clusterland 2014-09-22--2014-09-28)
-Jeremy
Hi Jeremy,
Thanks for the info.
I was accustomed to a small and probably less complex set of servers than Wikimedia has, and it is possible that the severs' acolytes were doing more than I knew about to keep the servers happy, but I notice that the number of Wikimedia Analytics trouble reports seems to be an order of magnitude greater than what I observed for network services in the place where I last worked in an IT department. I think the complexity and scale of the Wikimedia environment may be contributing factors, but I was wondering if there is more to the story, for example there may be inherently fragile tools used by WM Analytics. I am just curious.
Thanks,
Pine On Oct 12, 2014 6:02 PM, "Jeremy Baron" jeremy@tuxmachine.com wrote:
On Oct 12, 2014 8:49 PM, "Pine W" wiki.pine@gmail.com wrote:
I've noticed in my time on this list that it seems like the reliability
of Wikimedia Analytics services is a bit spotty. I've worked in IT services, and our web and email servers' reliability seemed pretty good, comparable to the Wikimedia content delivery services.
Who is "our"?
Which services do you have in mind?
Some services (e.g. wikimetrics, reportcard) run on labs so they inherent the lower uptime expectations of labs vs. prod.
Some services were historically not very reliable but should soon be more reliable than they ever were before. (see footnote 1 at bottom of https://lists.wikimedia.org/pipermail/analytics/2014-October/002539.html ; subject=Adventures in Clusterland 2014-09-22--2014-09-28)
-Jeremy
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics
I'd echo Jeremy's question: what services?
On 13 October 2014 00:43, Pine W wiki.pine@gmail.com wrote:
Hi Jeremy,
Thanks for the info.
I was accustomed to a small and probably less complex set of servers than Wikimedia has, and it is possible that the severs' acolytes were doing more than I knew about to keep the servers happy, but I notice that the number of Wikimedia Analytics trouble reports seems to be an order of magnitude greater than what I observed for network services in the place where I last worked in an IT department. I think the complexity and scale of the Wikimedia environment may be contributing factors, but I was wondering if there is more to the story, for example there may be inherently fragile tools used by WM Analytics. I am just curious.
Thanks,
Pine On Oct 12, 2014 6:02 PM, "Jeremy Baron" jeremy@tuxmachine.com wrote:
On Oct 12, 2014 8:49 PM, "Pine W" wiki.pine@gmail.com wrote:
I've noticed in my time on this list that it seems like the reliability
of Wikimedia Analytics services is a bit spotty. I've worked in IT services, and our web and email servers' reliability seemed pretty good, comparable to the Wikimedia content delivery services.
Who is "our"?
Which services do you have in mind?
Some services (e.g. wikimetrics, reportcard) run on labs so they inherent the lower uptime expectations of labs vs. prod.
Some services were historically not very reliable but should soon be more reliable than they ever were before. (see footnote 1 at bottom of https://lists.wikimedia.org/pipermail/analytics/2014-October/002539.html ; subject=Adventures in Clusterland 2014-09-22--2014-09-28)
-Jeremy
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics
On Oct 12, 2014 9:02 PM, "Jeremy Baron" jeremy@tuxmachine.com wrote:
Some services (e.g. wikimetrics, reportcard) run on labs so they inherent
the lower uptime expectations of labs vs. prod.
s/inherent/inherit/
Hi Pine,
as usual, not speaking for the Foundation, just writing down my own 2 cents.
On Sun, Oct 12, 2014 at 05:49:11PM -0700, Pine W wrote:
I've noticed in my time on this list that it seems like the reliability of Wikimedia Analytics services is a bit spotty.
I can see that this impression might arise.
However, I think such an impression does not take into account that Analytics is typically sitting at the end of a longish pipeline with many ever-moving interdependencies. Email servers on the other hand are “rather” self-contained.
Also wikis stay up if Analytics services stops. So while we could make our Analytics infrastructure more robust in many places, I am not sure if that is money well spent in all cases. Having a dashboard lag for a day or two is not nice, but it's not breaking the wikis.
Finally, I think it's worth bringing the list's scope into play. The analytics list is read by the general public, but is also a way that we can reach out to people within the foundation that are using Analytics infrastructure.
Hence, more things get brought to the list than would surface on targeted lists or for closed down shops. Most of the issues we bring to the list might have impact for researchers or devs running one-off queries and what they might eventually read off of data. But they hardly are real breakage.
To put this into more context, let's have alook at the Incident documentation at
https://wikitech.wikimedia.org/wiki/Incident_documentation
. For 2014, we're owning only 3 items there. None of them had direct impact on our users.
Have fun, Christian
Christian Aistleitner, 13/10/2014 11:49:
Hence, more things get brought to the list than would surface on targeted lists or for closed down shops.
+1 I think recently information quality on this list has increased a lot. If one looks only at public mailing lists archives (as opposed to private mailing lists, RT, bugzilla, IRC, services themselves) it's easy to miss the complete picture.
Nemo
Understood, thank you Christian.
Pine
On Mon, Oct 13, 2014 at 2:49 AM, Christian Aistleitner < christian@quelltextlich.at> wrote:
Hi Pine,
as usual, not speaking for the Foundation, just writing down my own 2 cents.
On Sun, Oct 12, 2014 at 05:49:11PM -0700, Pine W wrote:
I've noticed in my time on this list that it seems like the reliability
of
Wikimedia Analytics services is a bit spotty.
I can see that this impression might arise.
However, I think such an impression does not take into account that Analytics is typically sitting at the end of a longish pipeline with many ever-moving interdependencies. Email servers on the other hand are “rather” self-contained.
Also wikis stay up if Analytics services stops. So while we could make our Analytics infrastructure more robust in many places, I am not sure if that is money well spent in all cases. Having a dashboard lag for a day or two is not nice, but it's not breaking the wikis.
Finally, I think it's worth bringing the list's scope into play. The analytics list is read by the general public, but is also a way that we can reach out to people within the foundation that are using Analytics infrastructure.
Hence, more things get brought to the list than would surface on targeted lists or for closed down shops. Most of the issues we bring to the list might have impact for researchers or devs running one-off queries and what they might eventually read off of data. But they hardly are real breakage.
To put this into more context, let's have alook at the Incident documentation at
https://wikitech.wikimedia.org/wiki/Incident_documentation
. For 2014, we're owning only 3 items there. None of them had direct impact on our users.
Have fun, Christian
-- ---- quelltextlich e.U. ---- \ ---- Christian Aistleitner ---- Companies' registry: 360296y in Linz Christian Aistleitner Kefermarkterstrasze 6a/3 Email: christian@quelltextlich.at 4293 Gutau, Austria Phone: +43 7946 / 20 5 81 Fax: +43 7946 / 20 5 81 Homepage: http://quelltextlich.at/
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics