Re: [Sks-devel] Adding DB_INIT_LOCK to sks-keyserver (revisited)

[Phil Pennock]

On 2010-02-24 at 12:46 -0500, Jeff Johnson wrote:
> The fundamental issue (imho) is that Berkeley DB does locks per page, not
> locks per record, and so its pretty easy to self-deadlock.

Not with a single thread of control.

Reading over that discussion (it pre-dates my list membership and I
wasn't as conscientious as you), I have one suspicion of the cause.

Locks are held by cursors.  If a cursor isn't closed properly, that
could cause a later operation to deadlock.  SKS uses cursors.

Is it plausible that there's a handle leak somewhere in the cursor code
which can lead to a cursor not being closed in a timely manner?  Perhaps
kept around and re-used, with implicit close?

http://www.oracle.com/technology/documentation/berkeley-db/db/programmer_reference/lock_notxn.html

-Phil