I'm not sure how long it was, but it was certainly some days. I'll continue to index once an hour as long as possible.
--> Eric On Aug 30, 11:04 am, Pat Allan <[email protected]> wrote: > HiEric > > I'm not sure about the script, but just wanted to make it clear that you > should be running a full index regularly (whether that's hourly or daily) - > otherwise the delta index grows to be large enough to make the delta indexing > so slow that there's no advantage there. > > Especially given this script is constantly adding changes, I'd recommend > keeping full indexing happening once an hour (or at least once a day) - when > you turned off the full indexing, how long did it take for you to see issues > with results not being returned? > > -- > Pat > > On 29/08/2011, at 5:08 PM,Ericwrote: > > > > > > > > > Thanks for the update Pat. We'll give 2.0.1 a try- cautiously ;) > > > My apps use MVA filtering extensively so the realtime index filtering > > is definitely a non-starter at this point. > > > I use sphinx with delta indexing and delayed job. We have upstart to > > restart both sphinx and delayed job if necessary. The data sets for > > my clients are small enough that I can run a full reindex every hour > > across each client server. > > > Are you aware of any issues that can arise with delta indices if you > > don't regularly run a full index? The one inexplicable problem I've > > had is if I don't reindex every hour- it seems that we eventually see > > inconsistencies where the delta indexing stops. > > > Also I run a functional check via a webservice on each server every 15 > > minutes to make sure sphinx is running. I basically update a record > > then try to search for the record with an attribute and confirm that I > > see the result. I also have a check to see if the # of results for > > several commonly used models is above a certain threshold. I had a > > production issue once where sphinx had been restarted with a new > > configuration file but not the application. Sphinx continued to work, > > but periodically returned zero results for a model that had records. > > It was hard to track down because it repro'd sporadically. > > > This is the monitoring routine that I run: > > > def run_test_on url > > http_response = '' > > begin > > ts_i = Time.now.to_i > > HTTPI.put("#{url}/sphinx_checks/1? > > sphinx_check[check_ts]=#{ts_i}") > > sleep(2) > > > response = HTTPI.get("#{url}/sphinx_checks.json? > > check_ts=#{ts_i}") > > http_response = response.body > > response_json = ActiveSupport::JSON.decode http_response > > updated_ts = response_json['SphinxCheck'].first['check_ts'] > > > unless updated_ts.to_s == ts_i.to_s > > raise Exception.new("Error trying to run update url=#{url}, > > http_response=#{http_response}, checked #{ts_i}") > > end > > > # Tell sphinx check to check any models that need checking > > response = HTTPI.get("#{url}/sphinx_checks/1.json? > > check_models=1") > > http_response = response.body > > response_json = ActiveSupport::JSON.decode http_response > > unless response_json.empty? > > raise Exception.new("Error trying to run check_models > > url=#{url}, got response=#{http_response} - at #{ts_i}") > > end > > rescue Exception => e > > HoptoadNotifier.notify( > > :error_message => "Got an error against #{url}: #{e.inspect}, > > http_response=#{http_response}", > > :backtrace => e.backtrace > > ) > > > end > > end > > > I'm going to add an insert/delete as well. > > > -->Eric > > > On Aug 28, 1:04 pm, Pat Allan <[email protected]> wrote: > >> HiEric > > >> I've used 2.0.1 locally (and it's supported in recent releases of > >> TS/Riddle), but not yet on production. I haven't heard of it being > >> unstable, mind you, so I'd recommend giving it a shot. I'm pretty sure > >> others are using it without issues. > > >> As for realtime indices, TS doesn't currently support that. It requires > >> significant changes, and I've started toying with a rewrite, but that's > >> not going to appear in the near future. The realtime indices are missing > >> some pretty key features anyway - such as prefix/infix indexing and > >> possibly morphologies/stemming and charset tables. Multi-value attributes > >> for realtime indices aren't part of the 2.0.1 release either (they are > >> supported in recent commits though). > > >> So, all said - I reckon give 2.0.1 a shot, but you won't get the benefits > >> of realtime indices any time soon. > > >> -- > >> Pat > > >> On 28/08/2011, at 8:52 PM,Ericwrote: > > >>> Have folks migrated to Sphinx 2.0.1 yet? I'm wondering if anybody has > >>> production experience. I've been using Sphinx 0.9.9-release (r2117) > >>> with good success for some time now but never saw a more recent > >>> production release. > > >>> I know there are some new realtime indexing features that look very > >>> attractive as an alternative to delta indices, also wondering if there > >>> is a reasonable way to use that with thinking sphinx. > > >>> Thanks! > >>>Eric > > >>> -- > >>> 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 > >>> athttp://groups.google.com/group/thinking-sphinx?hl=en. > > > -- > > 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 > > athttp://groups.google.com/group/thinking-sphinx?hl=en. -- 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.
