On Mon, Aug 10, 2015 at 10:05 AM, Dave Page <dpage@pgadmin.org> wrote:
On Fri, Aug 7, 2015 at 7:20 PM, Stefan Kaltenbrunner
<stefan@kaltenbrunner.cc> wrote: > On 08/07/2015 11:11 AM, Stefan Kaltenbrunner wrote: >> On 08/07/2015 10:12 AM, Dave Page wrote: >>> On Thu, Aug 6, 2015 at 8:27 PM, Stefan Kaltenbrunner >>> <stefan@kaltenbrunner.cc> wrote: >>>> On 08/06/2015 09:01 PM, Josh Berkus wrote: >>>>> Folks: >>>>> >>>>> If you search: >>>>> >>>>> http://www.postgresql.org/search/?u=%2Fdocs%2F9.5%2F&q=tablesample >>>>> >>>>> You turn up a number of doc pages detailing the new TABLESAMPLE >>>>> functionality. However, if you actually click on the links: >>>>> >>>>> http://www.postgresql.org/docs/9.5/static/tablesample-method-functions.html >>>>> >>>>> http://www.postgresql.org/docs/9.5/static/catalog-pg-tablesample-method.html >>>>> >>>>> >>>>> ... they are 404. Doc build errors? >>>> >>>> no this looks like an artifact of the alpha2 release - it seems like the >>>> docs loaded to the website for 9.5 are currently out-of-sync with the >>>> webspider feeding our search engine(in this particular case the reason >>>> is the TABLESAMPLE rewrite from tom). I assume dave has that on his >>>> radar... >>> >>> As far as I'm aware, it's fully automated, but I'm not at all familiar >>> with the code and am probably not going to be able to investigate >>> until next week due to other commitments. Not sure when Magnus is >>> back, but he can probably help with some pointers at least. >> >> well I think the point is the timing of the docs loading process during >> the release vs the timing of the job that does the spidering. >> Since I was not aware how far you had progressed with the release >> (especially when you had done the docs loading) it made not a lot of >> sense investigating further on the search/spider side. > > After the recent reindexing runs, search is now providing the correct urls. > While this is a very minor issue usually (because it is usually > extremely rare that we make changes to the docs that shift urls around > especially in tagged/stable branches) maybe we should consider adding a > forced search reindex after a docs push to our release notes?
Not a bad idea. Do we have a simple way of forcing it at the moment?
Go to selek and run the "cron_websites" script manually (/home/pgsearch/cron_websites - run as user pgsearch). Needs to be done after the docs loaded obviously.
For now we should probably just put that in the step-by-step instructions. Since it runs on a completely separate box, it's not trivial to integrate it into the docloads scripts -- which would be a nicer long-term solution of course.