Bugs item #3512872, was opened at 2012-03-29 09:26 Message generated for change (Comment added) made by shcheklein You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=713730&aid=3512872&group_id=129076
Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: General >Group: Sedna (current&development) >Status: Closed >Resolution: Invalid Priority: 8 Private: No Submitted By: Michael Woytowitz (zazz-coder) Assigned to: Alexander Kalinin (allexk) Summary: Readonly query is not querying current state Initial Comment: I initially discovered this issue when using the Java API (jar). But the same issue exists when run in the command line client (se_term) Given a document in the database. If execute and update on that doc (in \upd mode) and view the doc the current version is retrieved. but if switch to readonly mode (\ro) and query the same document the result is the version of the doc as it existed at DB startup. When in readonly mode the query should still retrieve the current version of the document. I expect that \ro mode to only 'not lock for update' the document (or doc in a collection) and thus allow other connections to update the document while a read is performed. I encountered this issue because I am attempting to reduce multi-user dead locking - I am using the Java API and Sedna jar. I desire to issue queries (of the current state of the DB) in read only mode and then switch to update mode when performing updates while using the same active connection. in the case of se_term, It seems the readonly is looking at the data as it existed when the DB if first started (not the current state). The Java API is exhibiting this same behavior. ---------------------------------------------------------------------- >Comment By: Ivan Shcheklein (shcheklein) Date: 2012-03-31 07:02 Message: That's how read-only transactions work. You can try to tune upd_crt value. For details refer: - http://sedna.org/progguide/ProgGuidesu2.html#x5-130001.2.3 (SEDNA_ATTR_CONCURRENCY_TYPE) - http://sedna.org/adminguide/AdminGuidesu3.html#x7-100002.3.1 (upd-crt parameter) For brief description of upd_crt parameter see here: - http://thread.gmane.org/gmane.text.xml.sedna/1166/focus=1169 - https://sourceforge.net/mailarchive/forum.php?thread_name=4D65570C.2030809%40gmail.com&forum_name=sedna-discussion ---------------------------------------------------------------------- Comment By: Michael Woytowitz (zazz-coder) Date: 2012-03-30 12:08 Message: Hi, Could you please clarify why this is not a bug? Seems to me the current behavior does not follow what would be typically expected. Meaning, I would expect to retrieve the same query result regardless of whether I'm in \ro mode or \upd mode. I would only expect this mode to affect updates. Can you explain what the use case if that would result in the current behavior when switching from \ro to \upd and back to \ro in se_term. Thank you ---------------------------------------------------------------------- Comment By: Ilya Taranov (mozinsur) Date: 2012-03-30 06:36 Message: That is not a bug actually ---------------------------------------------------------------------- Comment By: Michael Woytowitz (zazz-coder) Date: 2012-03-29 09:31 Message: correction to paragraph 3: ....and thus allow multiple connections to read the current document without locking the document. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=713730&aid=3512872&group_id=129076 ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Sedna-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sedna-discussion
