Hi Numan, Thank you very much, I had pushed the patch new version https://github.com/openvswitch/ovs/pull/178
Thanks Zhengwei [email protected] From: Numan Siddique Date: 2017-05-22 15:33 To: 高正伟 CC: Ben Pfaff; ovs dev Subject: Re: Re: [ovs-dev] [PATCH] Supporting ovn-northd service HA depend on OVNDB-HA On Fri, May 19, 2017 at 7:21 AM, 高正伟 <[email protected]> wrote: Thanks Numan I think that supporting OVNDB and ovn-northd HA depend on one pacemaker is an idea, In case of thinking OVN-HA, user will not individually care about OVNDB or ovn-northd HA. Hi Zhangwei, I do agree with you. But what I am trying to say is to make it configurable. I tested with your patch again with tripleo and it is having few issues. 1. Since you are not passing "--ovn-manage-ovsdb=no", when "ovn-ctl stop_northd" is called, it also stops ovsdb-servers. We probably don't want to stop ovsdb-servers unless, pacemaker calls "stop" action. 2. When a master node is demoted, ovn-northd is not stopped since "ovn-ctl stop_northd" code is not hit. I tested with the below modifications and applying this patch - https://patchwork.ozlabs.org/patch/765224/. It is working as expected. Can you please test on your side and resubmit another version if it is fine with you. If the user wants to manage ovn_northd using the ocf resources, he could pass manage_northd=yes while creating the resource. -----------------------cut here ------------------------------------- diff --git a/ovn/utilities/ovndb-servers.ocf b/ovn/utilities/ovndb-servers.ocf index 40c5541..018d904 100755 --- a/ovn/utilities/ovndb-servers.ocf +++ b/ovn/utilities/ovndb-servers.ocf @@ -7,6 +7,7 @@ : ${NB_MASTER_PROTO_DEFAULT="tcp"} : ${SB_MASTER_PORT_DEFAULT="6642"} : ${SB_MASTER_PROTO_DEFAULT="tcp"} +: ${MANAGE_NORTHD_DEFAULT="no"} CRM_MASTER="${HA_SBIN_DIR}/crm_master -l reboot" CRM_ATTR_REPL_INFO="${HA_SBIN_DIR}/crm_attribute --type crm_config --name OVN_REPL_INFO -s ovn_ovsdb_master_server" OVN_CTL=${OCF_RESKEY_ovn_ctl:-${OVN_CTL_DEFAULT}} @@ -15,6 +16,7 @@ NB_MASTER_PORT=${OCF_RESKEY_nb_master_port:-${NB_MASTER_PORT_DEFAULT}} NB_MASTER_PROTO=${OCF_RESKEY_nb_master_protocol:-${NB_MASTER_PROTO_DEFAULT}} SB_MASTER_PORT=${OCF_RESKEY_sb_master_port:-${SB_MASTER_PORT_DEFAULT}} SB_MASTER_PROTO=${OCF_RESKEY_sb_master_protocol:-${SB_MASTER_PROTO_DEFAULT}} +MANAGE_NORTHD=${OCF_RESKEY_manage_northd:-${MANAGE_NORTHD_DEFAULT}} # Invalid IP address is an address that can never exist in the network, as # mentioned in rfc-5737. The ovsdb servers connects to this IP address till @@ -90,6 +92,14 @@ ovsdb_server_metadata() { <content type="string" /> </parameter> + <parameter name="manage_northd" unique="1"> + <longdesc lang="en"> + If set to yes, manages ovn-northd service. ovn-northd will be started in + the master node. + </longdesc> + <shortdesc lang="en">manage ovn-northd service</shortdesc> + <content type="string" /> + </parameter> </parameters> <actions> @@ -122,12 +132,17 @@ ovsdb_server_notify() { # the right thing at startup ocf_log debug "ovndb_server: $host_name is the master" ${CRM_ATTR_REPL_INFO} -v "$host_name" - # Startup ovn-northd service - ${OVN_CTL} start_northd + if [ "$MANAGE_NORTHD" = "yes" ]; then + # Startup ovn-northd service + ${OVN_CTL} --ovn-manage-ovsdb=no start_northd + fi else - # Stop ovn-northd service - ${OVN_CTL} stop_northd + if [ "$MANAGE_NORTHD" = "yes" ]; then + # Stop ovn-northd service. Set --ovn-manage-ovsdb=no so that + # ovn-ctl doesn't stop ovsdb-servers. + ${OVN_CTL} --ovn-manage-ovsdb=no stop_northd + fi # Synchronize with the new master ocf_log debug "ovndb_server: Connecting to the new master ${OCF_RESKEY_CRM_meta_notify_promote_uname}" ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${MASTER_IP} \ @@ -289,6 +304,13 @@ ovsdb_server_start() { } ovsdb_server_stop() { + if [ "$MANAGE_NORTHD" = "yes" ]; then + # Stop ovn-northd service in case it was running. This is required + # when the master is demoted. For other cases, it would be a no-op. + # Set --ovn-manage-ovsdb=no so that ovn-ctl doesn't stop ovsdb-servers. + ${OVN_CTL} --ovn-manage-ovsdb=no stop_northd + fi + ovsdb_server_check_status case $? in $OCF_NOT_RUNNING) return ${OCF_SUCCESS};; -------------------------------------------------------------------------- Thanks Numan Thanks. 在 2017-05-18 20:53:22，"Numan Siddique" <[email protected]> 写道： On Wed, May 17, 2017 at 9:15 PM, Ben Pfaff <[email protected]> wrote: From: Zhengwei Gao <[email protected]> As ovn-northd servcie parse network element between ovnnb_db and ovnsb_db, it ensures ovn-northd service connecting to ovnnb_db and ovnsb_db. OVNDB-HA was supported with pacemaker, ovn-northd service will be failover fllowing OVNDB-HA. --- This was posted as a github pull request: https://github.com/openvswitch/ovs/pull/176 ovn/utilities/ovndb-servers.ocf | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/ovn/utilities/ovndb-servers.ocf b/ovn/utilities/ovndb-servers.ocf index 908cb3c17d84..40c55411e828 100755 --- a/ovn/utilities/ovndb-servers.ocf +++ b/ovn/utilities/ovndb-servers.ocf @@ -122,8 +122,12 @@ ovsdb_server_notify() { # the right thing at startup ocf_log debug "ovndb_server: $host_name is the master" ${CRM_ATTR_REPL_INFO} -v "$host_name" + # Startup ovn-northd service + ${OVN_CTL} start_northd else + # Stop ovn-northd service + ${OVN_CTL} stop_northd # Synchronize with the new master ocf_log debug "ovndb_server: Connecting to the new master ${OCF_RESKEY_CRM_meta_notify_promote_uname}" ${OVN_CTL} demote_ovnnb --db-nb-sync-from-addr=${MASTER_IP} \ I tried this patch and it works. In the case of OpenStack and Tripleo, puppet is used to create and manage the pacemaker resources for OVN [1] and it also starts ovn-northd as a pacemaker service with colocation constraint set to the virtual ip created for the OVN DB servers. I think it makes more sense to start ovn-northd based on a flag (like START_NORTHD) so that the user or the configuration systems can chose. Thanks Numan [1] - https://review.openstack.org/#/c/372274/10/manifests/profile/pacemaker/ovn_northd.pp -- 2.10.2 _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
