Thanks for the tips! Upgrading Sphinx does indeed fix my problem, at least in development mode (will be curious to see if we have issues on our production server, which does 10k+ queries/day). Now, on to the rest of my problems :)
On Mar 18, 4:53 am, Anibal Cucco <[email protected]> wrote: > Hi Bill, > > You need to upgrade your sphinx version. Install Sphinx 0.9.8.1-release. > This is the link where this problem is > discussed:http://groups.google.com/group/thinking-sphinx/browse_thread/thread/c... > > Besides that, take a look to what Morten sent because we found some other > problems with datetime deltas and that link can help you in the future. > > -Anibal > > On Wed, Mar 18, 2009 at 8:39 AM, Morten <[email protected]> wrote: > > > Hi Bill, > > > There are some records on the sphinx merge issues here: > >http://www.sphinxsearch.com/bugs/my_view_page.php > > > I didn't realize what the problem was before doing things "by hand" as > > some error messages got swallowed. I know that work has been done in > > that area since, so I'm not sure what the symptoms would look like > > today - if they are at all merge issues. > > > Br, > > > Morten > > > On Mar 18, 10:33 am, wbharding <[email protected]> wrote: > > > I'd be interested in hearing more about the symptoms of these > > > "unreliable sphinx merges?" I have just started testing using the > > > datetime delta, and at the end of the delta update, my indexes are > > > consistently broken (i.e., before the delta update I can do searches, > > > after the delta completes, all queries come up empty. They don't give > > > any error). > > > > Don't know if it matters, but the .spa file for my index is not > > > updated after the merge, whereas all the other index files are > > > updated. > > > > Getting this working -- along with a datetime delta that could > > > consider child dependencies when deciding what to reindex -- would > > > solve most of the important challenges we're facing with TS as we > > > scale. > > > > For the record, the code I'm using to try to get the delta working is > > > straight vanilla: > > > > set_property :delta => :datetime, :threshold => 1.day > > > > and > > > > rake thinking_sphinx:index:delta > > > > This is running in development mode with Mongrel, no special options > > > setup in the sphinx.yml. Running Sphinx 0.9.8-release (r1371). It > > > looks like the index is working, in that it takes a long time to > > > process and it finds results proportional to the number of changed > > > records. The "only problem" is that the indexes are broken and > > > unusable afterwards... :) > > > > Bill > > > > On Feb 16, 12:49 am, Pat Allan <[email protected]> wrote: > > > > > Hi Morten > > > > > I think there's definitely room for anotherdeltamethod - delayed but > > > > with a manual rake call at regular intervals, likedatetime(and > > > > hopefully the thinking_sphinx:index:deltatask can be re-used). At > > > > this point it doesn't seem like merging is always reliable, so go for > > > > it. > > > > > Cheers > > > > > -- > > > > Pat > > > > > On 14/02/2009, at 3:42 AM, Morten wrote: > > > > > > Hey. I've been trying out variousdeltaindexing mechanisms on a > > > > > moderately busy site. These are my findings: > > > > > > :delta=> true, calls out to indexer immediately, not good when busy > > > > > :delta=> :datetime, relies on sphinx merge which is unstable > > > > > :delta=> :delayed, creates ~2 jobs each time a change is made, lots > > > > > of overhead > > > > > > I think :delta=> :datetimeis the "correct" solution for me, but I > > > > > cannot rely on sphinx to do things right here. So I'm looking for a > > > > > 4thdeltamechanism that doesn't make 1 save equal to 1 or more > > > > > indexer calls and searchd SIGHUP's. > > > > > > I'm considering implementing a :delta=> :bulk option which would be a > > > > > cross between :delta=> true and :delta=> :delayed > > > > > > What do people do in general on busy sites? Am I misinterpreting the > > > > > currentdeltapossibilities or is there a reliable way to do bulk > > > > >deltaupdating currently? > > > > > > Thanks. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Thinking Sphinx" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/thinking-sphinx?hl=en -~----------~----~----~----~------~----~------~--~---
