I think I may have seen Sup peg IO R/W in the way you saw, but when I
restarted Sup with debugging, it went away. Do let me know if you see it
again; I think there's an infinite loop somewhere.
Edward
Excerpts from Edward Z. Yang's message of Mon Sep 03 14:09:57 -0400 2012:
> OK, cracked a fix; you need this extra patch:
>
> commit f9ea07f3c4982ab46d8171fdba8eabc3af00c840
> Author: Edward Z. Yang <ezy...@mit.edu>
> Date: Mon Sep 3 14:09:34 2012 -0400
>
> sync_back after writing to index, not before.
>
> Signed-off-by: Edward Z. Yang <ezy...@mit.edu>
>
> diff --git a/lib/sup/index.rb b/lib/sup/index.rb
> index 13798d6..4b533a7 100644
> --- a/lib/sup/index.rb
> +++ b/lib/sup/index.rb
> @@ -657,10 +657,6 @@ EOS
> end
>
> def sync_message m, overwrite
> - ## TODO: we should not save the message if the sync_back failed
> - ## since it would overwrite the location field
> - m.sync_back
> -
> doc = synchronize { find_doc(m.id) }
> existed = doc != nil
> doc ||= Xapian::Document.new
> @@ -703,6 +699,10 @@ EOS
> @xapian.replace_document docid, doc
> end
>
> + # sync_back must be after label update, so that inotify gets
> + # fresh data from the index
> + m.sync_back
> +
> m.labels.each { |l| LabelManager << l }
> true
> end
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
