Hello,
Many thanks for your suggestion. I was not aware of the session table problem.
I managed to circumvent the cache problem (not a real solution).
Was able to find a local backup (rsync) including the var/cache and var/news/cache directories with megabytes of data and done at a day with normal behaviour.
Uploaded that old cache data and now the site is running at normal speed.
Some pages were displaying at old cached versions (!) from that date, despite admin interface showing their new versions.
Well, an edit publish cycle refreshed the cache content for these pages. Not the best solution, but worked.
From inside the admin panel, flushing the cache did not solve.
Weird. Seems as ezp not being able to create cache directories or files. but it can update content and recreate cache files already there.
It is a trick, and a real solution should be investigated.
regards. Andre Felipe
---
A Debian user never dies. Issues a last command:
shutdown -h now