any way to clear a huge update_seq?

[Michal Suchanek <su...@kovagroup.ca>]

I am running couchdb 0.11 on mac os x in a development environment. I have
about 50,000 documents which I have migrated from a mysql db. In the course
of migration, I have deleted and reimported this data about 15 times and now
the update_seq on the database is around 900,000. Now, when I add new views
(or modify current ones), the build takes forever as it checkpoints against
all of the previous (now-deleted) revisions of each document. I get nearly
1M lines of "checkpointing view update at seq 132299 for"... (where the seq
goes to nearly 900k).

I am not replicating this db at this time and have no need for old revision
stubs etc. I just want to be able to build views against existing documents
and discard the old revision data (and metadata)!

I have tried db compaction, view cleanup, and view compaction several times.
Nothing seems to make a difference and new views seem to checkpoint at every
or every second update_seq. This takes quite a while.

What am I missing?
-- 
View this message in context: http://couchdb-development.1959287.n2.nabble.com/any-way-to-clear-a-huge-update-seq-tp5237014p5237014.html
Sent from the CouchDB Development mailing list archive at Nabble.com.

Re: any way to clear a huge update_seq?

[Sebastian Cohnen <se...@googlemail.com>]

Hey Michal,

you don't want to reset the update_seq. You want to get rid of old documents. The only way to achieve this, is to set _revs_limit and run compaction.

http://wiki.apache.org/couchdb/HTTP_database_API#Accessing_Database-specific_options


best
Sebastian

On 30.06.2010, at 03:21, Michal Suchanek wrote:

> I am running couchdb 0.11 on mac os x in a development environment. I have
> about 50,000 documents which I have migrated from a mysql db. In the course
> of migration, I have deleted and reimported this data about 15 times and now
> the update_seq on the database is around 900,000. Now, when I add new views
> (or modify current ones), the build takes forever as it checkpoints against
> all of the previous (now-deleted) revisions of each document. I get nearly
> 1M lines of "checkpointing view update at seq 132299 for"... (where the seq
> goes to nearly 900k).
> 
> I am not replicating this db at this time and have no need for old revision
> stubs etc. I just want to be able to build views against existing documents
> and discard the old revision data (and metadata)!
> 
> I have tried db compaction, view cleanup, and view compaction several times.
> Nothing seems to make a difference and new views seem to checkpoint at every
> or every second update_seq. This takes quite a while.
> 
> What am I missing?
> -- 
> View this message in context: http://couchdb-development.1959287.n2.nabble.com/any-way-to-clear-a-huge-update-seq-tp5237014p5237014.html
> Sent from the CouchDB Development mailing list archive at Nabble.com.