在 2023/5/4 18:14, Ilya Maximets 写道:
On 5/4/23 12:12, Ilya Maximets wrote:
On 4/11/23 11:07, 汪翰林 wrote:
在 2023/4/7 21:20, Simon Horman 写道:
On Fri, Apr 07, 2023 at 02:59:04PM +0800, wanghanlin wrote:
When 'cs' is not connected to a server, then
ovsdb_cs_db_compose_lock_request__ will update
'cs_db' lock flag, but the lock flag will not
be updated next.
Signed-off-by: wanghanlin <wanghan...@corp.netease.com>
Hi,
Are there also error conditions where this situation can arise?
This happens when reconnect_disable is called.
Hi. Why exactly this condition is bad? While composing the
lock request the 'has_lock' flag will be dropped to 'false'.
The lock request will be created, but the send will fail, since
there is no connection. However, this shouldn't have any
consequences, because we do not claim to have a lock. Once we
re-connect a new request should be formed and sent out. Or am
I missing something?
Even worse, if we do not drop the flag, the application will still
think that it has the lock, while it doesn't.
Best regards, Ilya Maximets.
In bridge_run, ovsdb_idl_has_lock will check the 'has_lock' flag and
return immediately when 'has_lock' is false.
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
