manifoldcf-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Erlend GarĂ¥sen <e.f.gara...@usit.uio.no>
Subject Re: Hop count problem
Date Wed, 14 Aug 2013 11:29:00 GMT
On 8/13/13 4:16 PM, Karl Wright wrote:
> Hi Erlend,
>
> I see what must be happening.  The intrinsiclink table already has the
> link to the skuespill document in it, and because of that, nothing in
> the hopcount world is even getting looked at.  So in a nutshell, the
> problem is that somehow the hopcount table's data was messed up, but now
> there's no good way to recover.
>
> I would really like to know how it got messed up in the first place, but
> since there's been a lot of activity on that machine it would be a
> challenge to come up with the exact sequence of events.  If you think
> you remember it, please write it down and maybe try it on your test
> instance.  But for now, the simplest way to get the production instance
> back up and running is to do the following:

I'm sorry, but that will be very difficult since we had so many problems 
with this host and MCF itself at that time. The only thing I remember is 
that I changed the hop count model after the job had been created. That 
may be an issue ("Delete unreachable documents" to "Keep unreachable 
documents, forever")

> - Note all the job settings and configuration
> - Delete the job
> - Recreate the job
> - Run the job

Not a problem. The only thing is that MCF will delete those documents 
from Solr until I have a new job up and running again. Minor issue.

> Since there are very few documents in the job, it does not sound like
> much of a problem to do that.  Would this work for you?
> Karl

Thanks again, Karl. Yes, the remaining documents are being fetched at 
the time of writing.

Interesting about the theory regarding CONNECTORS-764. I may have 
changed the hop count settings after the job was created.

I will double-check our other jobs in case they are skipping similar 
documents as well and recreate them if necessary.

Erlend

Mime
View raw message