expiry-waaaaaaaaaaaaaaaait !!!

expiry-waaaaaaaaaaaaaaaait !!!

[Didier Verna]

	From time to time (I couldn't figure out the criteria), after reading
one or two new mails in an nnml folder (the summary contains only those new
ones), gnus take hours to expire the articles.

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /      E.N.S.T. INF C201.1      mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   (33) 01 45 81 73 46
                      75634 Paris  cedex 13     Fax.   (33) 01 45 81 31 19

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Lars Magne Ingebrigtsen]

Didier Verna <verna@inf.enst.fr> writes:

> 	From time to time (I couldn't figure out the criteria), after reading
> one or two new mails in an nnml folder (the summary contains only those new
> ones), gnus take hours to expire the articles.

Is the active range of this group wide?  Look in the nnml active
file. 

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Didier Verna]

Lars Magne Ingebrigtsen writes:

Lars> Is the active range of this group wide? Look in the nnml active file.

	It seems that I notice a real hangup for groups with a range of more
than, say 4000 articles. However, I've been experimenting this for only a few
betas I think.

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /      E.N.S.T. INF C201.1      mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   (33) 01 45 81 73 46
                      75634 Paris  cedex 13     Fax.   (33) 01 45 81 31 19

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Hrvoje Niksic]

Didier Verna <verna@inf.enst.fr> writes:

> From time to time (I couldn't figure out the criteria), after
> reading one or two new mails in an nnml folder (the summary contains
> only those new ones), gnus take hours to expire the articles.

Do you have old dormant articles in those groups?

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Didier Verna]

Hrvoje Niksic writes:

Hrvoje> Do you have old dormant articles in those groups?

	Hmmmm, yes, one or two. And some ticked ones also.

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /      E.N.S.T. INF C201.1      mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   (33) 01 45 81 73 46
                      75634 Paris  cedex 13     Fax.   (33) 01 45 81 31 19

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Hrvoje Niksic]

Didier Verna <verna@inf.enst.fr> writes:

> Hrvoje Niksic writes:
> 
> Hrvoje> Do you have old dormant articles in those groups?
> 
> Hmmmm, yes, one or two. And some ticked ones also.

Get rid of them.  The way total expiry in nnml works is that nnml
checks all the articles in the range.  Those old dormant/ticked
articles glue the lower endpoint of the range to a constant position,
while the upper endpoint keeps growing.

The bottom line is that when you exit the group, thousands of
non-existant files are stat'ted, which slows things down.  Either get
rid of the dormant/ticked messages, or use auto-expiry.

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Kai.Grossjohann]

Hrvoje Niksic <hniksic@srce.hr> writes:

  > Get rid of them.  The way total expiry in nnml works is that nnml
  > checks all the articles in the range.  Those old dormant/ticked
  > articles glue the lower endpoint of the range to a constant position,
  > while the upper endpoint keeps growing.

Another alternative is to, every once in a while, move all articles
from the group to the same group.  This will renumber the messages
such that they have consecutive numbers again.  Of course, the gap
will soon begin to grow again, so this must be repeated periodically.

I use auto-expire to prevent this, as Hrvoje suggested.

kai
-- 
I like _\bb_\bo_\bt_\bh kinds of music.

Re: expiry-waaaaaaaaaaaaaaaait !!!

[Hrvoje Niksic]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> I use auto-expire to prevent this, as Hrvoje suggested.

Just for the record: I don't.  I much prefer auto-expiry to
total-expiry, and I simply take care not to have old dormant/ticked
articles lying around.

The correct fix for this problem has been proposed by David Moore some
time ago: instead of a single "range" of articles, use a list of
ranges.  This should cure the problem.