"dev" <dev-boun...@openvswitch.org> wrote on 03/31/2016 02:21:34 PM:
> From: "Ansari, Shad" <shad.ans...@hpe.com> > To: "dev@openvswitch.org" <dev@openvswitch.org> > Date: 03/31/2016 02:21 PM > Subject: [ovs-dev] [PATCH 1/1] ovsdb-idl: Retain column values of > deleted rows until change-track is cleared. > Sent by: "dev" <dev-boun...@openvswitch.org> > > (This is in reference to mail thread - http://openvswitch.org/ > pipermail/dev/2016-March/068331.html) > > --- > > ovsdb-idl: Retain column values of deleted rows until change-track is cleared. > > When change tracking is enabled, only the row uuid of deleted rows > is available. > The column values are unparsed and the ovsdb_datum values are cleared so they > are not available for inspection. This change leaves the column values around > for inspection. The column values are cleared (unparsed) upon clearing the > change track. > > Signed-off-by: Shad Ansari <shad.ans...@hpe.com> I freely admit that I haven't tested this, but that my experience with change tracking while working on the incremental processing patches makes me question its usefulness. My experience with trying to clear the change track and re-establish it leads to history being lost [this is likely from my not fully understanding what happens with sequence numbers over turning tracking on and off] and so it is more efficient to just leave change tracking on. Under this scenario, I'm pretty convinced that this code would amount to a memory leak. Ryan _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
