Index keeps changing even though no update or delete

  1. Stop mongrel server

  2. Delete the index directory

  3. ruby script/console production
    Event.rebuild_index

  4. Get the list of items in the index directory.
    _8rs.cfs
    segments
    segments_8rt

  5. Start mongrel server

  6. click on various links to show event details information

  7. After 4 minutes get the list of items in the index directory
    _8rs.cfs
    _8rs_0.del
    _8rt.cfs
    ferret-write.lck
    segments
    segments_8rv

8)In the last 4 minutes no update was done for any event. No event was
deleted.

  1. Error I get is:
    A Ferret::FileNotFoundError occurred in eventinfo#index:

File Not Found Error occured at <except.c>:117 in xpop_context
Error occured in fs_store.c:329 - fs_open_input
tried to open
“/home/username/apps/eii_production/shared/ferret_production/_8rs_1.del”
but it doesn’t exist:

  1. I get the above error because the list of items in the index keeps
    changing every couple of minutes even though no update/delete operation
    is beine done.

Question:

  1. Since no update/delete operation was done I am assuming that the list
    of items in the index directory should not change. Is the assumption
    correct?
  2. It is my understanding that when segments are merged, internally
    everything is being tracked. Then how come some of the event throw error
    because “_8rs_1.del” is no more there.

Thanks

Forgot to mention that I am using Ferret 0.11.4 and the most stable
version of AAF on a linux machine.

I was running two mongrel instances behind pound. Still no update/delete
was done so that shouldn’t cause any problem in my view.

Raj S. wrote:

I was running two mongrel instances behind pound. Still no update/delete
was done so that shouldn’t cause any problem in my view.

Did you ever get this figured out? I’m having the same problem running
two Mongrels behind an Apache proxy. I suspect it might have something
to do with one Mongrel deleting another Mongrel’s indexes somehow.

Incidentally, I wasn’t able to follow the

  1. ruby script/console production
    Event.rebuild_index

Step, as I get an “Event is nil” error when I try to run that line.

Bill Harding wrote:

Raj S. wrote:

I was running two mongrel instances behind pound. Still no update/delete
was done so that shouldn’t cause any problem in my view.

Did you ever get this figured out? I’m having the same problem running
two Mongrels behind an Apache proxy. I suspect it might have something
to do with one Mongrel deleting another Mongrel’s indexes somehow.

Incidentally, I wasn’t able to follow the

  1. ruby script/console production
    Event.rebuild_index

Step, as I get an “Event is nil” error when I try to run that line.

The problem with my setup seems to have been umm…everything.

The first issue was that in our development environment we didn’t need
to run the ferret server to access ferret results, but in production (at
least with multiple Mongrels) we do need to run it. This was resolved
by following the instructions on the acts_as_ferret wiki (Install:
http://projects.jkraemer.net/acts_as_ferret/wiki. Use:
http://projects.jkraemer.net/acts_as_ferret/wiki/DrbServer) for setting
up the ferret server.

My auxiliary problem was that I had somehow gotten a version of the
plugin in my [app root]/vendor/plugins folder, which was missing the
ferret_server.rb that had been in the real install location (off the
Ruby directory). Once I discovered this, I copied the ferret server
stuff into my [app root]/script directory, as the front page of the
acts_as_ferret wiki instructs.

My final problem was that I wasn’t putting my ferret indexes in the
Capistrano shared directory. But now I am all better
(Ferret and capistrano, how to keep the indexes? - Ferret - Ruby-Forum).

Bill