Bad news, I am afraid. MySQL seems to always put null values at the front of the index, and that cannot be changed through any means I can find. This is different from all other databases I know of.
The only possible fixes for the problem are as follows: (1) Not use a null doc priority but instead use an actual special number that is guaranteed to always sort to the end. This would be non-trivial because this column is a FLOAT value, and round-off errors will prevent the ManifoldCF code from reliably using a special number like that on all databases. (2) Use docpriorities that are ordered in the opposite way - which would work ONLY for MySQL and would break all other databases. I'll create a ticket and think about the problem some more. Karl On Mon, Jan 21, 2013 at 8:48 AM, Karl Wright <[email protected]> wrote: > Hi Shigeki, > > I reviewed the code in detail. At the time CONNECTORS-290 was fixed, > all document priorities were set to null whenever a job was paused or > aborted, so what I suspected might be the problem cannot in fact > happen. > > The most likely possible explanation for MySQL's behavior, therefore, > is that MySQL orders null docpriority values BEFORE all other rows in > the index it is using for queue stuffing. I have no other way of > explaining why it thinks it needs to go through 6.5 million rows > before it gets to the ones that are active. > > If this is the case, it may be possible to tell MySQL to order null > column values to the END instead of the beginning of the index. I'll > do some research on this later and get back to you. > > Thanks, > Karl > > > > On Mon, Jan 21, 2013 at 6:21 AM, Karl Wright <[email protected]> wrote: >> Are there any large paused or aborted jobs present on the same >> ManifoldCF? If so, can you tell me whether the job is paused, or >> aborted? (I am betting paused...) >> >> Karl >> >> On Mon, Jan 21, 2013 at 5:59 AM, Shigeki Kobayashi >> <[email protected]> wrote: >>> Hi Karl, >>> >>> >>> Here is the explain. There isn't such sort... >>> >>> mysql> explain SELECT >>> t0.id,t0.jobid,t0.dochash,t0.docid,t0.status,t0.failtime,t0.failcount,t0.priorityset >>> FROM jobqueue t0 FORCE INDEX (i1358228295210) WHERE t0 IN ('P','G') AND >>> t0.checkaction='R' AND t0.checktime<=1358649661663 AND EXISTS(SELECT 'x' >>> FROM jobs t1 WHERE t1.status IN ('A','a') AND t1.id=t0.jobid AND >>> t1.priority=5) AND NOT EXISTS(SELECT 'x' FROM jobqueue t2 WHERE >>> t2.dochash=t0.dochash AND t2.status IN ('A','F','a','f','D','d') AND >>> t2.jobid!=t0.jobid) AND NOT EXISTS(SELECT 'x' FROM prereqevents t3,events t4 >>> WHERE t0.id=t3.owner AND t3.eventname=t4.name) ORDER BY t0.docpriority ASC >>> LIMIT 4800; >>> +----+--------------------+-------+--------+----------------------------------------------+----------------+---------+-------------------------+------+-------------+ >>> | id | select_type | table | type | possible_keys >>> | key | key_len | ref | rows | Extra | >>> +----+--------------------+-------+--------+----------------------------------------------+----------------+---------+-------------------------+------+-------------+ >>> | 1 | PRIMARY | t0 | index | NULL >>> | I1358228295210 | 25 | NULL | 4800 | Using where | >>> | 4 | DEPENDENT SUBQUERY | t3 | ref | I1358228295216 >>> | I1358228295216 | 8 | manifoldcf.t0.id | 1 | | >>> | 4 | DEPENDENT SUBQUERY | t4 | eq_ref | PRIMARY >>> | PRIMARY | 767 | manifoldcf.t3.eventname | 1 | Using index | >>> | 3 | DEPENDENT SUBQUERY | t2 | ref | >>> I1358228295209,I1358228295212,I1358228295211 | I1358228295209 | 122 | >>> manifoldcf.t0.dochash | 1 | Using where | >>> | 2 | DEPENDENT SUBQUERY | t1 | eq_ref | PRIMARY,I1358228295219 >>> | PRIMARY | 8 | manifoldcf.t0.jobid | 1 | Using where | >>> +----+--------------------+-------+--------+----------------------------------------------+----------------+---------+-------------------------+------+-------------+ >>> 5 rows in set (0.00 sec) >>> >>> >>> Regards, >>> >>> >>> Shigeki >>> >>> >>> >>> 2013/1/21 Karl Wright <[email protected]> >>>> >>>> takes too long >>>> MIME-Version: 1.0 >>>> Content-Type: multipart/alternative; boundary=14dae9399bd9676e5704d3c356e9 >>>> >>>> --14dae9399bd9676e5704d3c356e9 >>>> Content-Type: text/plain; charset="utf-8" >>>> Content-Transfer-Encoding: 7bit >>>> >>>> Can you get an EXPLAIN for this query? It sounds like it is >>>> disregarding the hint for some reason. >>>> >>>> Karl >>>> >>>> Sent from my Windows Phone >>>> From: Shigeki Kobayashi >>>> Sent: 1/20/2013 9:37 PM >>>> To: [email protected] >>>> Subject: Re: Crawling new/updated files using Windows share connection >>>> takes too long >>>> Hi Karl. >>>> >>>> I configured MySQL 5.5 to run MCF this time. >>>> The version of MCF is trunk 1.1dev downloaded on Dec, 12th. , which you >>>> fixed >>>> the slow query using "FORCE INDEX". Solr is 4.0 >>>> >>>> I thought is was fixed but the log shows that the following are slow >>>> queries. >>>> ------------------------------------------------------------------- >>>> # Time: 130120 11:41:10 >>>> # User@Host: manifoldcf[manifoldcf] @ localhost [127.0.0.1] >>>> # Query_time: 8.761087 Lock_time: 0.000163 Rows_sent: 17 Rows_examined: >>>> 6365233 >>>> SET timestamp=1358649670; >>>> SELECT >>>> t0.id,t0.jobid,t0.dochash,t0.docid,t0.status,t0.failtime,t0.failcount,t0.priorityset >>>> FROM jobqueue t0 FORCE INDEX (i1358228295210) WHERE t0.status IN ('P','G') >>>> AND t0.checkaction='R' AND t0.checktime<=1358649661663 AND EXISTS(SELECT >>>> 'x' FROM jobs t1 WHERE t1.status IN ('A','a') AND t1.id=t0.jobid AND >>>> t1.priority=5) AND NOT EXISTS(SELECT 'x' FROM jobqueue t2 WHERE >>>> t2.dochash=t0.dochash AND t2.status IN ('A','F','a','f','D','d') AND >>>> t2.jobid!=t0.jobid) AND NOT EXISTS(SELECT 'x' FROM prereqevents t3,events >>>> t4 WHERE t0.id=t3.owner AND t3.eventname=t4.name) ORDER BY t0.docpriority >>>> ASC LIMIT 4800; >>>> >>>> # Time: 130120 11:41:18 >>>> # User@Host: manifoldcf[manifoldcf] @ localhost [127.0.0.1] >>>> # Query_time: 7.714277 Lock_time: 0.000123 Rows_sent: 0 Rows_examined: >>>> 6365182 >>>> SET timestamp=1358649678; >>>> SELECT docpriority,jobid,dochash,docid FROM jobqueue t0 FORCE INDEX >>>> (i1358228295210) WHERE status IN ('P','G') AND checkaction='R' AND >>>> checktime<=1358649661663 AND EXISTS(SELECT 'x' FROM jobs t1 WHERE >>>> t1.status >>>> IN ('A','a') AND t1.id=t0.jobid) ORDER BY docpriority ASC LIMIT 1; >>>> >>>> Regards, >>>> >>>> >>>> Shigeki >>>> >>>> >>>> >>>> 2013/1/18 Karl Wright <[email protected]> >>>> >>>> > Hi Shigeki, >>>> > >>>> > What database is ManifoldCF configured to use in this case? Do you >>>> > see any indication of slow queries in the ManifoldCF log? >>>> > >>>> > >>>> > Karl >>>> > >>>> > On Fri, Jan 18, 2013 at 5:27 AM, Shigeki Kobayashi >>>> > <[email protected]> wrote: >>>> > > Hello >>>> > > >>>> > > >>>> > > I would like some advice to improve crawling time of new/updated files >>>> > using >>>> > > Windows share connection. >>>> > > >>>> > > I crawl file in Windows server and index them into Solr. >>>> > > >>>> > > Currently, the second crawling of two hundred thousands files takes >>>> > over 5 >>>> > > hours, even though any files are not updated, created, deleted. >>>> > > >>>> > > I assume MCF does the following processes (let me know if I am wrong) >>>> > > >>>> > > - obtain updated time of a file >>>> > > - compare the updated time with the one MCF obtained last time >>>> > > crawling( >>>> > > probably stored in DB) >>>> > > - if they are different MCF recognizes the file is to be indexed. >>>> > > >>>> > > If the above processes are done for two thousands files, what part of >>>> > > the >>>> > > processes could take time the most? obtaining updated time? reading >>>> > > data >>>> > > from DB? what could be done to increase the crawling time do you >>>> > > think? >>>> > > >>>> > > Please give me some advice. >>>> > > >>>> > > >>>> > > Regards, >>>> > > >>>> > > Shigeki >>>> > > >>>> > > >>>> > >>>> >>>> --14dae9399bd9676e5704d3c356e9 >>>> Content-Type: text/html; charset="utf-8" >>>> Content-Transfer-Encoding: quoted-printable >>>> >>>> <html><head><meta content=3D"text/html; charset=3Dutf-8" >>>> http-equiv=3D"Cont= >>>> ent-Type"></head><body><div><div style=3D"font-family: Calibri,sans-serif; >>>> = >>>> font-size: 11pt;"><br>Can you get an EXPLAIN for this query? It >>>> sound= >>>> s like it is disregarding the hint for some >>>> reason.<br><br>Karl<br><br>Sent= >>>> from my Windows Phone<br></div></div><hr><span style=3D"font-family: >>>> Tahom= >>>> a,sans-serif; font-size: 10pt; font-weight: bold;">From: </span><span >>>> style= >>>> =3D"font-family: Tahoma,sans-serif; font-size: 10pt;">Shigeki >>>> Kobayashi</sp= >>>> an><br><span style=3D"font-family: Tahoma,sans-serif; font-size: 10pt; >>>> font= >>>> -weight: bold;">Sent: </span><span style=3D"font-family: >>>> Tahoma,sans-serif;= >>>> font-size: 10pt;">1/20/2013 9:37 PM</span><br><span style=3D"font-family: >>>> = >>>> Tahoma,sans-serif; font-size: 10pt; font-weight: bold;">To: </span><span >>>> st= >>>> yle=3D"font-family: Tahoma,sans-serif; font-size: >>>> 10pt;">[email protected]= >>>> ache.org</span><br><span style=3D"font-family: Tahoma,sans-serif; >>>> font-size= >>>> : 10pt; font-weight: bold;">Subject: </span><span style=3D"font-family: >>>> Tah= >>>> oma,sans-serif; font-size: 10pt;">Re: Crawling new/updated files using >>>> Wind= >>>> ows share connection takes too long</span><br><br></body></html><div >>>> dir=3D= >>>> "ltr">Hi Karl.<div><br></div><div>I configured MySQL 5.5 to run MCF this >>>> ti= >>>> me.</div><div>The version of MCF is trunk 1.1dev downloaded on Dec, 12th. >>>> ,= >>>> which you fixed</div><div style>the slow query using "FORCE >>>> INDEX&quo= >>>> t;. Solr is 4.0</div> >>>> >>>> <div style><br></div><div style>I thought is was fixed but the log shows >>>> th= >>>> at =C2=A0the following are slow queries.=C2=A0</div><div><div >>>> class=3D"gmai= >>>> >>>> l_extra">------------------------------------------------------------------= >>>> -</div> >>>> <div class=3D"gmail_extra"> >>>> <div class=3D"gmail_extra"># Time: 130120 11:41:10</div><div >>>> class=3D"gmail= >>>> _extra"># User@Host: manifoldcf[manifoldcf] @ localhost >>>> [127.0.0.1]</div><d= >>>> iv class=3D"gmail_extra"># Query_time: 8.761087 =C2=A0Lock_time: 0.000163 >>>> R= >>>> ows_sent: 17 =C2=A0Rows_examined: 6365233</div> >>>> >>>> <div class=3D"gmail_extra">SET timestamp=3D1358649670;</div><div >>>> class=3D"g= >>>> mail_extra">SELECT <a >>>> href=3D"http://t0.id";>t0.id</a>,t0.jobid,t0.dochash,t= >>>> 0.docid,t0.status,t0.failtime,t0.failcount,t0.priorityset FROM jobqueue t0 >>>> = >>>> FORCE INDEX (i1358228295210) WHERE t0.status IN ('P','G') >>>> A= >>>> ND t0.checkaction=3D'R' AND t0.checktime<=3D1358649661663 AND >>>> EX= >>>> ISTS(SELECT 'x' FROM jobs t1 WHERE t1.status IN >>>> ('A','a= >>>> ') AND <a href=3D"http://t1.id";>t1.id</a>=3Dt0.jobid AND >>>> t1.priority=3D= >>>> 5) AND NOT EXISTS(SELECT 'x' FROM jobqueue t2 WHERE >>>> t2.dochash=3Dt0= >>>> .dochash AND t2.status IN >>>> ('A','F','a','f',= >>>> 'D','d') AND t2.jobid!=3Dt0.jobid) AND NOT EXISTS(SELECT >>>> &#= >>>> 39;x' FROM prereqevents t3,events t4 WHERE <a >>>> href=3D"http://t0.id";>t0.= >>>> id</a>=3Dt3.owner AND t3.eventname=3D<a >>>> href=3D"http://t4.name";>t4.name</a>= >>>> ) ORDER BY t0.docpriority ASC LIMIT 4800;</div> >>>> >>>> <div class=3D"gmail_extra"><br></div><div class=3D"gmail_extra"># Time: >>>> 130= >>>> 120 11:41:18<br></div><div class=3D"gmail_extra"># User@Host: >>>> manifoldcf[ma= >>>> nifoldcf] @ localhost [127.0.0.1]<br></div><div class=3D"gmail_extra"># >>>> Que= >>>> ry_time: 7.714277 =C2=A0Lock_time: 0.000123 Rows_sent: 0 >>>> =C2=A0Rows_examine= >>>> d: 6365182</div> >>>> >>>> <div class=3D"gmail_extra">SET timestamp=3D1358649678;</div><div >>>> class=3D"g= >>>> mail_extra">SELECT docpriority,jobid,dochash,docid FROM jobqueue t0 FORCE >>>> I= >>>> NDEX (i1358228295210) WHERE status IN ('P','G') AND >>>> checkac= >>>> tion=3D'R' AND checktime<=3D1358649661663 AND EXISTS(SELECT >>>> '= >>>> ;x' FROM jobs t1 WHERE t1.status IN ('A','a') AND <a >>>> hr= >>>> ef=3D"http://t1.id";>t1.id</a>=3Dt0.jobid) =C2=A0ORDER BY docpriority ASC >>>> LI= >>>> MIT 1;</div> >>>> >>>> <div><br></div><div class=3D"gmail_extra" style>Regards,</div><div >>>> class=3D= >>>> "gmail_extra"><br></div><div class=3D"gmail_extra"><br></div><div >>>> class=3D"= >>>> gmail_extra" style>Shigeki</div><div class=3D"gmail_extra"><br></div><div >>>> c= >>>> lass=3D"gmail_extra"> >>>> >>>> <br></div><br><div class=3D"gmail_quote">2013/1/18 Karl Wright <span >>>> dir=3D= >>>> "ltr"><<a href=3D"mailto:[email protected]"; >>>> target=3D"_blank">daddywri@= >>>> gmail.com</a>></span><br><blockquote class=3D"gmail_quote" >>>> style=3D"marg= >>>> in:0px 0px 0px >>>> 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,20= >>>> 4);border-left-style:solid;padding-left:1ex"> >>>> >>>> Hi Shigeki,<br> >>>> <br> >>>> What database is ManifoldCF configured to use in this case? =C2=A0Do >>>> you<br= >>>> > >>>> see any indication of slow queries in the ManifoldCF log?<br> >>>> <br> >>>> <br> >>>> Karl<br> >>>> <br> >>>> On Fri, Jan 18, 2013 at 5:27 AM, Shigeki Kobayashi<br> >>>> <<a >>>> href=3D"mailto:[email protected]";>shigeki.kobayash= >>>> [email protected]</a>> wrote:<br> >>>> > Hello<br> >>>> ><br> >>>> ><br> >>>> > I would like some advice to improve crawling time of new/updated >>>> files= >>>> using<br> >>>> > Windows share connection.<br> >>>> ><br> >>>> > I crawl file in Windows server and index them into Solr.<br> >>>> ><br> >>>> > Currently, the second crawling of two hundred thousands files takes = >>>> =C2=A0over 5<br> >>>> > hours, even though any files are not updated, created, deleted.<br> >>>> ><br> >>>> > I assume MCF does the following processes (let me know if I am >>>> wrong)<= >>>> br> >>>> ><br> >>>> > - obtain updated time of a file<br> >>>> > - compare the updated time with the one MCF obtained last time >>>> crawlin= >>>> g(<br> >>>> > probably stored in DB)<br> >>>> > - if they are different MCF recognizes the file is to be indexed.<br> >>>> ><br> >>>> > If the above processes are done for two thousands files, what part of >>>> = >>>> the<br> >>>> > processes could take time the most? obtaining updated time? reading >>>> da= >>>> ta<br> >>>> > from DB? what could be done to increase the crawling time do you >>>> think= >>>> ?<br> >>>> ><br> >>>> > Please give me some advice.<br> >>>> ><br> >>>> ><br> >>>> > Regards,<br> >>>> ><br> >>>> > Shigeki<br> >>>> ><br> >>>> ><br> >>>> </blockquote></div><br><br clear=3D"all"><div><br></div><br><div><font >>>> face= >>>> =3D"'ms gothic', monospace">=C2=A0</font></div><div><br></div> >>>> </div></div></div> >>>> >>>> --14dae9399bd9676e5704d3c356e9-- >>> >>> >>> >>> >>>
