incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Victor Nicollet <vnicol...@runorg.com>
Subject Re: View row with doc==null
Date Wed, 16 Nov 2011 07:17:05 GMT
I have recently encountered this problem as well on a 1.1.0 production
database. Like you say, the "fix" was to rebuild the view. I do not know
the reason for this behavior, but I believe it would be important to
investigate.

On 16 November 2011 08:13, Marcello Nuccio <marcello.nuccio@gmail.com>wrote:

> No, the row is:
>
>   {"id":"510427_2011_1","key":["510427",2011,1],"value":1,"doc":null}
>
> Marcello
>
> 2011/11/16 Mark Hahn <mark@hahnca.com>:
> > Does this doc by any chance have an empty _id field?  If so that is a
> known
> > problem.  It cannot be fixed in a db when it happens.  Couch is supposed
> to
> > have a fix soon, if not already.
> >
> > On Tue, Nov 15, 2011 at 10:57 PM, Marcello Nuccio <
> marcello.nuccio@gmail.com
> >> wrote:
> >
> >> I've reloaded Futon at least three times...
> >>
> >> However, the problem is that the row with doc==null is still there,
> >> and the only way to get rid of it is to rebuild the view. I will try
> >> to find how to reproduce it in a new database.
> >>
> >> If anyone is interested to see the database, I can send it. It does
> >> not contain top secret data, and it's only a couple of MB.
> >>
> >> I have updated my program to ignore those rows. It is a good thing to
> >> do since it can happen if querying with stale=ok.
> >>
> >> Marcello
> >>
> >> 2011/11/15 Mark Hahn <mark@hahnca.com>:
> >> > You have to reload twice to update futon.  It is a pita.
> >> >
> >> > On Tue, Nov 15, 2011 at 5:32 AM, Marcello Nuccio
> >> > <marcello.nuccio@gmail.com>wrote:
> >> >
> >> >> 2011/11/15 Marcello Nuccio <marcello.nuccio@gmail.com>:
> >> >> > I've done some more testing:
> >> >> >
> >> >> > 1. I've restored the original db and view files (before compaction
> and
> >> >> cleanup)
> >> >> > 2. I've made a clone of the db and view with new name
> >> >> > 3. Verified that everything works as expected, i.e. both db return
> >> >> > identical results and all files are identical (done using curl
and
> >> >> > cmp).
> >> >> >
> >> >> > Here the tests:
> >> >> >
> >> >> > 1. Compacting the clone the row with doc==null disappears.
> >> >> > 2. Compacting the master the row is still here.
> >> >>
> >> >> Ops sorry, I did those two tests with Futon and forcing reload did
> not
> >> >> show the row in test 1. After clearing the cache completely, the row
> >> >> with doc==null is here.
> >> >>
> >> >> I verified with curl and both dbs return the row with doc==null.
> >> >>
> >> >> The only way I have found to delete it, is to delete the view file.
> >> >>
> >> >> Marcello
> >> >>
> >> >>
> >> >> > 3. Comparing with cmp *.couch and *.view, they are identical.
> >> >> > 4. Restarting couchdb does not help.
> >> >> >
> >> >> > What am I missing?
> >> >> >
> >> >> > Marcello
> >> >> >
> >> >> > 2011/11/15 Robert Newson <rnewson@apache.org>:
> >> >> >> That does sound wrong but not sure what steps to suggest next.
> >> >> >>
> >> >> >> B.
> >> >> >>
> >> >> >> On 15 November 2011 12:22, Marcello Nuccio <
> >> marcello.nuccio@gmail.com>
> >> >> wrote:
> >> >> >>> No.
> >> >> >>>
> >> >> >>> Marcello
> >> >> >>>
> >> >> >>> 2011/11/15 Robert Newson <rnewson@apache.org>:
> >> >> >>>> Did you query with stale=ok?
> >> >> >>>>
> >> >> >>>> B.
> >> >> >>>>
> >> >> >>>> On 15 November 2011 11:10, Marcello Nuccio <
> >> marcello.nuccio@gmail.com>
> >> >> wrote:
> >> >> >>>>> Don't know how, but I have a view that returns
the following
> row:
> >> >> >>>>>
> >> >> >>>>>
> >> >>  {"id":"510427_2011_1","key":["510427",2011,1],"value":1,"doc":null}
> >> >> >>>>>
> >> >> >>>>> i.e. it has doc==null when queried with include_docs=true.
The
> doc
> >> >> >>>>> with the given _id is deleted. The database is
compacted. The
> >> views
> >> >> >>>>> are compacted and cleaned (done via Futon). CouchDB
is version
> >> 1.1.1
> >> >> >>>>> built with build-couchdb on Ubuntu 11.10.
> >> >> >>>>>
> >> >> >>>>> Is this normal behaviour?
> >> >> >>>>>
> >> >> >>>>> Thanks,
> >> >> >>>>>  Marcello
> >> >> >>>>>
> >> >> >>>>
> >> >> >>>
> >> >> >>
> >> >> >
> >> >>
> >> >
> >>
> >
>



-- 
Victor Nicollet, Directeur Technique, www.runorg.com

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message