On 4/13/26 4:18 PM, Felix Huettner wrote: > Am Tue, Apr 07, 2026 at 10:45:03PM +0200 schrieb Ilya Maximets: >> On 3/17/26 9:28 AM, Felix Huettner via dev wrote: >> > This allows us to figure out how often we receive updates from upstream >> > dbs. In addition we can see e.g. during a reconnection if we did clear >> > the database (e.g. because monitor_cond_since did not work). >> > >> > Signed-off-by: Felix Huettner <[email protected]> >> > --- >> > lib/ovsdb-cs.c | 12 ++++++++++++ >> > 1 file changed, 12 insertions(+) >> > >> > diff --git a/lib/ovsdb-cs.c b/lib/ovsdb-cs.c >> > index df33a835d..0f0f88ef7 100644 >> > --- a/lib/ovsdb-cs.c >> > +++ b/lib/ovsdb-cs.c >> > @@ -20,6 +20,7 @@ >> > >> > #include <errno.h> >> > >> > +#include "coverage.h" >> > #include "hash.h" >> > #include "jsonrpc.h" >> > #include "openvswitch/dynamic-string.h" >> > @@ -38,6 +39,9 @@ >> > #include "uuid.h" >> > >> > VLOG_DEFINE_THIS_MODULE(ovsdb_cs); > > Hi Ilya, > > thanks for the review. > >> >> nit: An empty line here. > > done. > >> >> > +COVERAGE_DEFINE(ovsdb_cs_update); >> > +COVERAGE_DEFINE(ovsdb_cs_update_with_clear); >> > +COVERAGE_DEFINE(ovsdb_cs_update_from_monitor); >> >> Maybe drop the 'ovsdb_' prefix from these to make them a little sorter? > > done. > >> >> Also, have you considered putting these into idl code that parses events >> and count all types of events at the same location? Not sure if it's >> necessary, but might be a reasonable thing to do. > > The benefit i see in the cs code is that is then also directly usable > for relays.
Makes sense. > But if you prefer i can also split it and create two sets of counters. One for > the IDL and one for the relays. Let's keep in cs. I suppose, we could extend the ovsdb_cs_db_add_event() in the future to count all types of events, but it's not necessary for this patch. > > Thanks, > Felix > >> >> Best regards, Ilya Maximets. > > _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
