Hmm, not sure why the retries would behave any differently, to be honest. server would still be nil, and it'd just repeat five times before raising an error anyway, right? But yes, that's the source of 'Server not set'.
It could be related to the 'No response from searchd' errors - this comes down to the daemon itself not responding. Could be worth seeing how people have worked around that issue on the Sphinx forum, perhaps? Also, I was reading the other day of people using HAProxy in front of a few Sphinx daemons as a load balancer - perhaps that's worth looking into? https://github.com/freelancing-god/thinking-sphinx/issues/308 -- Pat On 04/01/2012, at 4:56 PM, Ngan wrote: > Looking a little deeper at the code, and specifically your change to > Riddle (https://github.com/freelancing-god/riddle/commit/ > 0fd884f1ae17e49257d1230383662148e8d1b96f). I think the difference is > that in the old code, if server was nil, it would still try to load > the TCPSocket: > TCPSocket.new(nil, 1234) > Which raises Errno::ECONNREFUSED, and gets caught by the rescue. I'm > guessing in the during the retries, one of the other if statements > evals to true. I think this explains the "Server not set" errors. > > Would this also explain the "No response from searched" errors too? > > > On Jan 3, 8:09 am, Ngan <[email protected]> wrote: >> I'm on TS 1.4.7, riddle 1.4.0. >> >> I guess my question would be what did you mean by "the errors around >> this weren't completely obvious?" >> Also, what are the main reasons a server would fail? It seems to me >> that the failures are often temporary that retrying all servers again >> X times wouldn't hurt. I guess I need to better understand what's >> happening that's causing these failures to better suggest a solution. >> I just can't see why so often BOTH servers are considered "failing." >> Any thoughts? >> >> On Jan 3, 12:00 am, "Pat Allan" <[email protected]> wrote: >> >> >> >> >> >> >> >>> What Riddle does is when the first server fails, it goes to the next, and >>> tries that. It repeats that until its out of servers, and then raises an >>> error. In 1.5.0 and earlier, the errors around this weren't completely >>> obvious. 1.5.1 has improved that - but it's surprising that it was all fine >>> before 1.5.1 for you. >> >>> -- >>> Pat >> >>> On 03/01/2012, at 6:58 PM, Ngan wrote: >> >>>> Also, not sure if it'll help, but I know that the errors does not >>>> happen on every single request, maybe every 5th. When I load a page >>>> for the first time (sometimes) it will error. A refresh will load >>>> fine. Also only notice this on production only as well... But that >>>> might be production has the high loads to cause these issue. >> >>>> On Jan 2, 11:55 pm, Ngan <[email protected]> wrote: >>>>> Sphinx 0.9.9 >> >>>>> sphinx.yml: >> >>>>> development: >>>>> delayed_job_priority: 9 >>>>> max_matches: 100000 >>>>> morphology: stem_en >>>>> timeout: 0.5 >>>>> port: &port 3312 >>>>> # If you wish to use multiple addresses (for failover), uncomment >>>>> the >>>>> # following lines: >>>>> # address: >>>>> # -111.111.111.111 >>>>> # -222.222.222.222 >>>>> # listen: *port >>>>> test: >>>>> delayed_job_priority: 9 >>>>> max_matches: 100000 >>>>> morphology: stem_en >>>>> port: *port >>>>> production: >>>>> delayed_job_priority: 9 >>>>> max_matches: 100000 >>>>> morphology: stem_en >>>>> timeout: 5.0 >>>>> port: &port 3312 >>>>> address: >>>>> - x.x.x.x >>>>> - x.x.x.x >>>>> listen: *port >>>>> bin_path: '/usr/local/bin' >> >>>>> On Jan 2, 6:12 pm, "Pat Allan" <[email protected]> wrote: >> >>>>>> What's your config/sphinx.yml looking like? What version of Sphinx are >>>>>> you using? >> >>>>>> -- >>>>>> Pat >> >>>>>> On 03/01/2012, at 5:53 AM, Ngan wrote: >> >>>>>>> I just upgraded to TS 1.4.11 and Riddle 1.5.1 and am now seeing a lot >>>>>>> of errors from Riddle: >> >>>>>>> "Server not set." and "No response from searched (status: , >>>>>>> version: )" >> >>>>>>> Any idea one what could have caused this? >> >>>>>>> -- >>>>>>> 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. > -- 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.
