[thredds] Joinexisting aggregation cache not hitting/rebuilding

• To: "thredds@xxxxxxxxxxxxxxxx" <thredds@xxxxxxxxxxxxxxxx>
• Subject: [thredds] Joinexisting aggregation cache not hitting/rebuilding
• From: "Blodgett, David" <dblodgett@xxxxxxxx>
• Date: Thu, 22 Aug 2013 11:31:25 -0500

THREDDS list,

I've got some aggregated services that use a scan element with no
recheckEvery attribute. The public face for these is here:
http://cida.usgs.gov/thredds/rfc_qpe/qpe_catalog.html

The aggregation cache files successfully write to the
${CATALINA_HOME}/content/thredds/cache/agg
directory using default threddsconfig.xml settings (no dir element set)

Today is August 20th. Some of the cache files in there are from August
12th. If I hit the OPeNDAP .html page for a dataset (where the cache file
is from August 12th) it hangs for a long time, and the time stamp of today
ends up on the cache file after it finally comes back.

Now if I refresh the OPeNDAP page, the time stamp on the cache file resets
to now, but the page comes back quickly, presumable not rebuilding the
cache, just touching it, or something. (Note that I've rebuilt the caches
for all these so they come back fast today, you won't be able to reproduce
the slowness)

Am I missing some configuration? Is this expected behavior, ie. the cache
files are rebuilt after they sit for some amount of time? I can verify that
restarting the instance doesn't cause the complete rescan / super slow
response.

A couple other things I've tried,
- I set the dir element to ${CATALINA_HOME}/content/thredds/cache/agg
- I symlinked ${CATALINA_HOME}/content/thredds/cache/agg to ${TOMCAT_HOME}/
content/thredds/cache/agg
No change from either of these.

Thanks for any guidance you can give on this!

Dave