I believe the condition was that the first row in the ROW range query does not exist in the database *and* the data to be returned existed in a CellStore (not the CellCache). If you were to re-load your data in 0.9.7.13 and then compact all of your tables with COMPACT RANGES ALL; then the query should exhibit the problematic behavior. Upgrading to 0.9.7.14 should fix the problem.
- Doug On Tue, Dec 17, 2013 at 9:51 PM, Xy Zheng <[email protected]> wrote: > timestamps be created using my insert client application. i used same APP > to insert data. so insert the same data twice and supply the same > timestamps both times. > > ps: what does particular condition mean? i don't understand clearly. > more detail explain will be appreciated. > > 在 2013年12月18日星期三UTC+8下午12时48分55秒，Doug Judd写道： >> >> I think the problem stemmed from a particular condition of your first >> database. That's why when you re-inserted the data, the problem went away, >> even for 0.9.7.13. >> >> If you insert the same data twice, Hypertable will deal with it in two >> different ways depending on whether or not you supply timestamps. If you >> supply the same timestamps both times you insert, Hypertable will handle it >> as if it was only inserted once. If you supply the data without >> timestamps, Hypertable will auto-assign timestamps, so you will get two >> copies of the data, each copy will have a different timestamp and you will >> see both versions when you query. >> >> - Doug >> >> >> On Tue, Dec 17, 2013 at 7:22 PM, Xy Zheng <[email protected]> wrote: >> >>> i sorry to omitted most important information. when I did test, the >>> data before stored in hypertable be deleted . >>> so i save data again to hypertable(v0.9.7.13) and test,record result. >>> then upgrade to 0.9.7.14.6a57db0 test and record result. >>> the data source two times saved to hypertable all same. >>> >>> by the way, is it the same data will be saved completely same in two >>> inserted to hypertable? >>> clearly i mean. the data saved to range will be changed or not? >>> thank you for giving me clearly explain. >>> >>> 在 2013年12月17日星期二UTC+8下午6时37分50秒，Doug Judd写道： >>>> >>>> Ok, if I'm reading the listing correctly, it now looks like both >>>> versions are working properly? Am I reading that correctly? >>>> >>>> - Doug >>>> >>>> >>>> On Tue, Dec 17, 2013 at 12:17 AM, Xy Zheng <[email protected]>wrote: >>>> >>>>> Hi Doug: >>>>> I just do what you said. and test result as appendix. >>>>> >>>>> in file >>>>> 1.capfile_0.9.7.13 means use your first supplyed >>>>> hyertable-v0.9.7.13. and >>>>> 2. capfile_0.9.7.14.6a57db0 means use 0.9.7.14.6a57db0 version you >>>>> given. >>>>> >>>>> >>>>> 在 2013年12月13日星期五UTC+8上午8时18分48秒，Doug Judd写道： >>>>>> >>>>>> I was able to reproduce the problem and have come up with a fix which >>>>>> you can find in the following package: >>>>>> >>>>>> http://www.hypertable.com/pub/hypertable-0.9.7.14.6a57db0-li >>>>>> nux-x86_64.tar.bz2 >>>>>> >>>>>> Please give this one a try and report back, letting us know if it >>>>>> does indeed fix the problem. Thanks! >>>>>> >>>>>> - Doug >>>>>> >>>>>> >>>>>> On Thu, Dec 12, 2013 at 7:37 AM, Doug Judd <[email protected]>wrote: >>>>>> >>>>>>> I think I know what's going on here. I will build you a fixed >>>>>>> package today. If you could give it a try and report back, I'd greatly >>>>>>> appreciate it. >>>>>>> >>>>>>> - Doug >>>>>>> >>>>>>> >>>>>>> On Thu, Dec 12, 2013 at 7:22 AM, Doug Judd <[email protected]>wrote: >>>>>>> >>>>>>>> Thanks. Are you certain that these log entries were gathered >>>>>>>> *after* you ran the problematic query? >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> On Thu, Dec 12, 2013 at 12:58 AM, <[email protected]> wrote: >>>>>>>> >>>>>>>>> Please check the attach file. >>>>>>>>> >>>>>>>>> >>>>>>>>> 在 2013年12月11日星期三UTC+8上午3时19分42秒，Doug Judd写道： >>>>>>>>>> >>>>>>>>>> Thanks. I've built an instrumented version of Hypertable which >>>>>>>>>> you can find here: >>>>>>>>>> >>>>>>>>>> http://www.hypertable.com/pub/hypertable-0.9.7.14.7266ca8-li >>>>>>>>>> nux-x86_64.tar.bz2 >>>>>>>>>> >>>>>>>>>> Please upgrade to this version and then start the cluster. Once >>>>>>>>>> the cluster is back up on the instrumented version, run the problem >>>>>>>>>> query: >>>>>>>>>> >>>>>>>>>> hypertable> select * from TAB_CDR_EVENT where >>>>>>>>>> '8795741566-51015A61BHE'<ROW<='D0000000000000000000zH ' and row >>>>>>>>>> regexp "B..$" limit 1; >>>>>>>>>> >>>>>>>>>> After the query has finished, grep for "issue1175" in the >>>>>>>>>> RangeServer logs and post the output. For example: >>>>>>>>>> >>>>>>>>>> cap shell >>>>>>>>>> cap> fgrep issue1175 /opt/hypertable/current/log/Hy >>>>>>>>>> pertable.RangeServer.log >>>>>>>>>> >>>>>>>>>> Thanks for your help! >>>>>>>>>> >>>>>>>>>> - Doug >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Mon, Dec 9, 2013 at 11:57 PM, <[email protected]> wrote: >>>>>>>>>> >>>>>>>>>>> Yes, 64 bit. tar is ok. >>>>>>>>>>> >>>>>>>>>>> It is OS information. >>>>>>>>>>> [cloudil@dlxa101 ~]$ uname -a >>>>>>>>>>> Linux dlxa101 2.6.32-279.el6.x86_64 #1 SMP Wed Jun 13 18:24:36 >>>>>>>>>>> EDT 2012 x86_64 x86_64 x86_64 GNU/Linux >>>>>>>>>>> >>>>>>>>>>> thanks >>>>>>>>>>> >>>>>>>>>>> On Tuesday, December 10, 2013 1:37:44 PM UTC+8, Doug Judd wrote: >>>>>>>>>>> >>>>>>>>>>>> That would be great, thanks! I will build you an instrumented >>>>>>>>>>>> version tomorrow and post a pointer to the package. In the >>>>>>>>>>>> meantime, try >>>>>>>>>>>> not to insert or delete any data from the table. What >>>>>>>>>>>> architecture and >>>>>>>>>>>> package type would you like me to build? 64-bit .tar.bz2? >>>>>>>>>>>> >>>>>>>>>>>> - Doug >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> On Mon, Dec 9, 2013 at 7:22 PM, <[email protected]> wrote: >>>>>>>>>>>> >>>>>>>>>>>>> Ok, it is my pleasure. Where can I get the instrumented >>>>>>>>>>>>> version? >>>>>>>>>>>>> >>>>>>>>>>>>> thanks >>>>>>>>>>>>> >>>>>>>>>>>>> 在 2013年12月10日星期二UTC+8上午3时57分53秒，Doug Judd写道： >>>>>>>>>>>>> >>>>>>>>>>>>>> Great, thank you. Would you be willing to run an >>>>>>>>>>>>>> instrumented version of Hypertable? It would help pin down >>>>>>>>>>>>>> exactly where >>>>>>>>>>>>>> the error is occurring. >>>>>>>>>>>>>> >>>>>>>>>>>>>> - Doug >>>>>>>>>>>>>> >>>>>>>>>>>>>> -- >>>>>>>>>>>>> You received this message because you are subscribed to the >>>>>>>>>>>>> Google Groups "Hypertable Development" group. >>>>>>>>>>>>> To unsubscribe from this group and stop receiving emails from >>>>>>>>>>>>> it, send an email to [email protected]. >>>>>>>>>>>>> To post to this group, send email to hyperta...@googlegroups. >>>>>>>>>>>>> com. >>>>>>>>>>>>> >>>>>>>>>>>>> Visit this group at http://groups.google.com/group >>>>>>>>>>>>> /hypertable-dev. >>>>>>>>>>>>> For more options, visit https://groups.google.com/grou >>>>>>>>>>>>> ps/opt_out. >>>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> -- >>>>>>>>>>>> Doug Judd >>>>>>>>>>>> CEO, Hypertable Inc. >>>>>>>>>>>> >>>>>>>>>>> -- >>>>>>>>>>> You received this message because you are subscribed to the >>>>>>>>>>> Google Groups "Hypertable Development" group. >>>>>>>>>>> To unsubscribe from this group and stop receiving emails from >>>>>>>>>>> it, send an email to [email protected]. >>>>>>>>>>> To post to this group, send email to [email protected] >>>>>>>>>>> . >>>>>>>>>>> Visit this group at http://groups.google.com/group >>>>>>>>>>> /hypertable-dev. >>>>>>>>>>> For more options, visit https://groups.google.com/groups/opt_out >>>>>>>>>>> . >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> -- >>>>>>>>>> Doug Judd >>>>>>>>>> CEO, Hypertable Inc. >>>>>>>>>> >>>>>>>>> -- >>>>>>>>> You received this message because you are subscribed to the Google >>>>>>>>> Groups "Hypertable Development" group. >>>>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>>>> send an email to [email protected]. >>>>>>>>> To post to this group, send email to [email protected]. >>>>>>>>> Visit this group at http://groups.google.com/group/hypertable-dev. >>>>>>>>> For more options, visit https://groups.google.com/groups/opt_out. >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>>>> Doug Judd >>>>>>>> CEO, Hypertable Inc. >>>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> Doug Judd >>>>>>> CEO, Hypertable Inc. >>>>>>> >>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> Doug Judd >>>>>> CEO, Hypertable Inc. >>>>>> >>>>> -- >>>>> You received this message because you are subscribed to the Google >>>>> Groups "Hypertable Development" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to [email protected]. >>>>> To post to this group, send email to [email protected]. >>>>> Visit this group at http://groups.google.com/group/hypertable-dev. >>>>> For more options, visit https://groups.google.com/groups/opt_out. >>>>> >>>> >>>> >>>> >>>> -- >>>> Doug Judd >>>> CEO, Hypertable Inc. >>>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "Hypertable Development" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To post to this group, send email to [email protected]. >>> Visit this group at http://groups.google.com/group/hypertable-dev. >>> For more options, visit https://groups.google.com/groups/opt_out. >>> >> >> >> >> -- >> Doug Judd >> CEO, Hypertable Inc. >> > -- > You received this message because you are subscribed to the Google Groups > "Hypertable Development" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/hypertable-dev. > For more options, visit https://groups.google.com/groups/opt_out. > -- Doug Judd CEO, Hypertable Inc. -- You received this message because you are subscribed to the Google Groups "Hypertable Development" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/hypertable-dev. For more options, visit https://groups.google.com/groups/opt_out.
