[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: back-bdb future

> -----Original Message-----
> From: Jonghyuk Choi [mailto:jongchoi@us.ibm.com]

> The recent index DB subdatabase patch of Howard
> seems to solve the deadlock problem ! (dbcache.c version 1.25)
> I experimented with 200K entries and the test run went well.
> I'll try a million entry run during the weekend.
> BTW, it does not seem to be related to the pthread mutex,
> because the deadlock occurred even when the number of threads
> is set to
> just 1.
> - Jong

Heh. I had a feeling that subdatabase stuff was a liability. Looking back at
your stack trace, I guess the dbcache lock wasn't involved at all. There's
only one thread that's actually waiting in libdb, so it has apparently
deadlocked with itself. Seems like we've hit a bug in BerkeleyDB, maybe you
should forward a copy of the stack traces to SleepyCat.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support