Re: ResultCache problems

[Golda Velez <gvelez@tucson.com>]

Christian is right - I had intended for people to run a cleanup script in
their crontab deleting entries older than a certain amount, but yes,
ResultCache should do its own cleanup when it runs.  Good suggestions
(below), and if we make the amount of time configurable in the install,
then sysadmins can tune it to their liking.   It seems to me we can
eliminate all race conditions if we just add the IP address of the
requestor to the name of the cache file.  Any objections to that?

--G

At 03:35 PM 9/16/99 -0400, Christian Vogler wrote:
>Hello folks,
>
>I learned the hard way yesterday that the result cache in the
>commercial version of webglimpse can eat up all space on the drive
>containing /tmp and never delete the offending files. The sponsors of
>our web site wouldn't be too happy with a denial of service attack, so
>I had to take it down.
>
>I think that this module should see some design changes to prevent
>this kind of thing from happening. It looks tricky, because web
>transcations are stateless. But maybe the follwing would be possible:
>
>- Limit the cache files to a specific amount of space, and
>  start deleting the oldest files when this space is exceeded.
>
>- Delete all cache files older than - say - 10 minutes.
>
>These two things could be done in the webglimpse script before it
>submits the query to glimpse. It also means that a query for the next
>page of hits could get the rug pulled out of from under it. Hence, I
>see the need for another change:
>
>- If the user requests the next page of hits and the result cache file
>  does not exist, webglimpse should resubmit the query to glimpse.
>
>We'd probably also have to be careful to avoid race conditions around
>removing the cache files when multiple users submit queries at the
>same time.
>
>Comments?
>
>- Christian
>
>
>
>
>
------------------------------------------------------------
Golda Velez         gvelez@tucson.com	        520-620-6878
Internet Workshop                          http://tucson.com
Webglimpse Search Software             http://webglimpse.net
		~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Help organize the world - index your own corner of the web