On 12/3/21 17:06, Dumitru Ceara wrote: > On 12/3/21 16:00, Terry Wilson wrote: >> On Fri, Dec 3, 2021 at 6:06 AM Dumitru Ceara <dce...@redhat.com> wrote: >>> >>> On 12/1/21 18:51, Terry Wilson wrote: >>>> Add support for monitor_cond_since / update3 to python-ovs to >>>> allow more efficient reconnections when connecting to clustered >>>> OVSDB servers. >>>> >>>> Signed-off-by: Terry Wilson <twil...@redhat.com> >>>> --- >>>> python/ovs/db/idl.py | 245 ++++++++++++++++++++++++++++++++++++------- >>>> tests/ovsdb-idl.at | 2 +- >>>> 2 files changed, 211 insertions(+), 36 deletions(-) >>>> >>> >>> Acked-by: Dumitru Ceara <dce...@redhat.com>
Thanks, Terry and Dumitru! I added a NEWS entry for that change and applied. I suppose, the next step here should be use of 'monitor_cond' for a monitor request to _Server database. Currently it uses simple 'monitor' that results in relatively bulky updates every time database "index" changes, and that happens on every update of a main database. C version uses 'monitor_cond' with a trivial condition in order to receive small 'update2' notifications. One problem I spotted that existed before this patch (so not a blocker) is that python-idl doesn't re-connect if it receives inconsistent update or a broken database schema. __parse_update() seems to trap all the parsing exceptions and only logs them. In case of schema or initial monitor reply parsing failures there is an extra check to force re-connection, but it's in the caller of __parse_update(), hence it can't actually get the exception. This, probably, need to be fixed. C version of a code uses ovsdb_cs_retry()/ovsdb_cs_flag_inconsistency() functions in similar cases. Best regards, Ilya Maximets. _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev
