On 02/04/2016 09:41 AM, Numan Siddique wrote:
> This reverts f20396e051ea4fd623bfc022cc78d9bd52a850e5
>
> f20396e.. was required to fix the failing ovn-sbctl testsuites.
> commit fb496f92ca1eeead8760b5cdfd857165f64deadf addresses these failing
> test cases in a proper way.
>
> With commit f20396e.. its possible that commit to the southbound db is
> delayed. When ovnnb_db_run is called, and if ctx->ovnsb_txn is NULL,
> ovnnb_db_run returns immediately without generating the logical flows.
> ovnnb_db_run is not called again until the northbound db seqno changes.
>
> Signed-off-by: Numan Siddique <nusid...@redhat.com>
> ---
> ovn/northd/ovn-northd.c | 14 ++------------
> 1 file changed, 2 insertions(+), 12 deletions(-)
>
> diff --git a/ovn/northd/ovn-northd.c b/ovn/northd/ovn-northd.c
> index 8083e25..e6271cf 100644
> --- a/ovn/northd/ovn-northd.c
> +++ b/ovn/northd/ovn-northd.c
> @@ -1877,7 +1877,6 @@ add_column_noalert(struct ovsdb_idl *idl,
> int
> main(int argc, char *argv[])
> {
> - unsigned int ovnnb_seqno, ovnsb_seqno;
> int res = EXIT_SUCCESS;
> struct unixctl_server *unixctl;
> int retval;
> @@ -1945,9 +1944,6 @@ main(int argc, char *argv[])
> add_column_noalert(ovnsb_idl_loop.idl, &sbrec_port_binding_col_mac);
> ovsdb_idl_add_column(ovnsb_idl_loop.idl,
> &sbrec_port_binding_col_chassis);
>
> - ovnnb_seqno = ovsdb_idl_get_seqno(ovnnb_idl_loop.idl);
> - ovnsb_seqno = ovsdb_idl_get_seqno(ovnsb_idl_loop.idl);
> -
> /* Main loop. */
> exiting = false;
> while (!exiting) {
> @@ -1958,14 +1954,8 @@ main(int argc, char *argv[])
> .ovnsb_txn = ovsdb_idl_loop_run(&ovnsb_idl_loop),
> };
>
> - if (ovnnb_seqno != ovsdb_idl_get_seqno(ctx.ovnnb_idl)) {
> - ovnnb_seqno = ovsdb_idl_get_seqno(ctx.ovnnb_idl);
> - ovnnb_db_run(&ctx);
> - }
> - if (ovnsb_seqno != ovsdb_idl_get_seqno(ctx.ovnsb_idl)) {
> - ovnsb_seqno = ovsdb_idl_get_seqno(ctx.ovnsb_idl);
> - ovnsb_db_run(&ctx);
> - }
> + ovnnb_db_run(&ctx);
> + ovnsb_db_run(&ctx);
>
> unixctl_server_run(unixctl);
> unixctl_server_wait(unixctl);
>
Thanks. On the surface this looks like a fine optimization, but I see
that it breaks things in some cases. I applied this patch to master.
--
Russell Bryant
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
