Early last week we started experiencing the following problem under
7.0.16 and 7.0.15
After a short period of usage, selects against a group of user
views seem to hang. In reality their process is CPU bound and
recursive calls for the session goes though the roof. A trace of
that session shows that they are in an endless loop executing the
following statement:
'select text from view$ where obj# = :1'
Killing the process and an instance restart seem to help as does
dropping and recreating the views, but 15 minutes to 1 hour later
the system is hosed again.
Oracle reported a bug in 7.0.13 that matches this description
(BUG 175425) but hasn't offered a solution. It is currently at
priority 1 and they are working on it.
We are copying the application to 7.1.3 in hopes that the bug doesn't
exist in that version but that process is not complete.
Has anyone seen something that looks like this? If so how did you
get around it? Comments?
