Propagate the shutdown event to the recursing ns_client(s)

Send the ns_query_cancel() on the recursing clients when we initiate the
named shutdown for faster shutdown.

When we are shutting down the resolver, we cancel all the outstanding
fetches, and the ISC_R_CANCEL events doesn't propagate to the ns_client
callback.

In the future, the better solution how to fix this would be to look at
the shutdown paths and let them all propagate from bottom (loopmgr) to
top (f.e. ns_client).

diff --git a/lib/ns/client.c b/lib/ns/client.c
index 95b2abed4038c1048d9bfb77246b665e1ceee8db..39ef9acd9f6314c7a7e9b2474f70437c2de395ec 100644 (file)
--- a/lib/ns/client.c
+++ b/lib/ns/client.c
@@ -120,10 +120,6 @@
 
 atomic_uint_fast64_t ns_client_requests = 0;
 
-static void
-clientmgr_attach(ns_clientmgr_t *source, ns_clientmgr_t **targetp);
-static void
-clientmgr_detach(ns_clientmgr_t **mp);
 static void
 clientmgr_destroy_cb(void *arg);
 static void
@@ -1680,7 +1676,7 @@ ns__client_put_cb(void *client0) {
 
        isc_mem_put(manager->mctx, client, sizeof(*client));
 
-       clientmgr_detach(&manager);
+       ns_clientmgr_detach(&manager);
 }
 
 /*
@@ -2319,7 +2315,7 @@ ns__client_setup(ns_client_t *client, ns_clientmgr_t *mgr, bool new) {
 
                *client = (ns_client_t){ .magic = 0 };
 
-               clientmgr_attach(mgr, &client->manager);
+               ns_clientmgr_attach(mgr, &client->manager);
 
                dns_message_create(client->manager->mctx,
                                   DNS_MESSAGE_INTENTPARSE, &client->message);
@@ -2381,7 +2377,7 @@ cleanup:
        }
 
        if (client->manager != NULL) {
-               clientmgr_detach(&client->manager);
+               ns_clientmgr_detach(&client->manager);
        }
 
        return (result);
@@ -2391,21 +2387,6 @@ cleanup:
  *** Client Manager
  ***/
 
-static void
-clientmgr_attach(ns_clientmgr_t *source, ns_clientmgr_t **targetp) {
-       int32_t oldrefs;
-
-       REQUIRE(VALID_MANAGER(source));
-       REQUIRE(targetp != NULL && *targetp == NULL);
-
-       oldrefs = isc_refcount_increment0(&source->references);
-       isc_log_write(ns_lctx, NS_LOGCATEGORY_CLIENT, NS_LOGMODULE_CLIENT,
-                     ISC_LOG_DEBUG(3), "clientmgr @%p attach: %d", source,
-                     oldrefs + 1);
-
-       *targetp = source;
-}
-
 static void
 clientmgr_destroy_cb(void *arg) {
        ns_clientmgr_t *manager = (ns_clientmgr_t *)arg;
@@ -2425,23 +2406,13 @@ clientmgr_destroy_cb(void *arg) {
 }
 
 static void
-clientmgr_detach(ns_clientmgr_t **mp) {
-       int32_t oldrefs;
-       ns_clientmgr_t *mgr = *mp;
-       *mp = NULL;
-
-       oldrefs = isc_refcount_decrement(&mgr->references);
-       isc_log_write(ns_lctx, NS_LOGCATEGORY_CLIENT, NS_LOGMODULE_CLIENT,
-                     ISC_LOG_DEBUG(3), "clientmgr @%p detach: %d", mgr,
-                     oldrefs - 1);
-       if (oldrefs == 1) {
-               isc_loop_t *loop = isc_loop_get(mgr->loopmgr, mgr->tid);
-
-               /* FIXME: Use isc_loopmgr_teardown() function instead? */
-               isc_async_run(loop, clientmgr_destroy_cb, mgr);
-       }
+clientmgr_destroy(ns_clientmgr_t *mgr) {
+       isc_loop_t *loop = isc_loop_get(mgr->loopmgr, mgr->tid);
+       isc_async_run(loop, clientmgr_destroy_cb, mgr);
 }
 
+ISC_REFCOUNT_IMPL(ns_clientmgr, clientmgr_destroy);
+
 isc_result_t
 ns_clientmgr_create(ns_server_t *sctx, isc_taskmgr_t *taskmgr,
                    isc_loopmgr_t *loopmgr, dns_aclenv_t *aclenv, int tid,
@@ -2485,13 +2456,20 @@ ns_clientmgr_create(ns_server_t *sctx, isc_taskmgr_t *taskmgr,
 }
 
 void
-ns_clientmgr_destroy(ns_clientmgr_t **managerp) {
-       REQUIRE(managerp != NULL);
-       REQUIRE(VALID_MANAGER(*managerp));
+ns_clientmgr_shutdown(ns_clientmgr_t *manager) {
+       ns_client_t *client;
+
+       REQUIRE(VALID_MANAGER(manager));
 
        MTRACE("destroy");
 
-       clientmgr_detach(managerp);
+       LOCK(&manager->reclock);
+       for (client = ISC_LIST_HEAD(manager->recursing); client != NULL;
+            client = ISC_LIST_NEXT(client, rlink))
+       {
+               ns_query_cancel(client);
+       }
+       UNLOCK(&manager->reclock);
 }
 
 isc_sockaddr_t *

diff --git a/lib/ns/include/ns/client.h b/lib/ns/include/ns/client.h
index 47b72685025db664a6e27cbb702f8af9f00d1db9..1f92486f61ac4bf4c7be40144439df1b7fc9e9d4 100644 (file)
--- a/lib/ns/include/ns/client.h
+++ b/lib/ns/include/ns/client.h
@@ -329,10 +329,10 @@ ns_clientmgr_create(ns_server_t *sctx, isc_taskmgr_t *taskmgr,
  */
 
 void
-ns_clientmgr_destroy(ns_clientmgr_t **managerp);
+ns_clientmgr_shutdown(ns_clientmgr_t *manager);
 /*%<
- * Destroy a client manager and all ns_client_t objects
- * managed by it.
+ * Shutdown a client manager and all ns_client_t objects
+ * managed by it
  */
 
 isc_sockaddr_t *
@@ -532,6 +532,8 @@ ns_client_findversion(ns_client_t *client, dns_db_t *db);
  * allocated by ns_client_newdbversion().
  */
 
+ISC_REFCOUNT_DECL(ns_clientmgr);
+
 isc_result_t
 ns__client_setup(ns_client_t *client, ns_clientmgr_t *manager, bool new);
 /*%<

diff --git a/lib/ns/interfacemgr.c b/lib/ns/interfacemgr.c
index 4f4253e488185bcc890c461ec8a466fdaf28eb75..dc86e2e9753a6726cd6a2276b9d62a8717c1b820 100644 (file)
--- a/lib/ns/interfacemgr.c
+++ b/lib/ns/interfacemgr.c
@@ -382,7 +382,7 @@ ns_interfacemgr_destroy(ns_interfacemgr_t *mgr) {
        clearlistenon(mgr);
        isc_mutex_destroy(&mgr->lock);
        for (size_t i = 0; i < mgr->ncpus; i++) {
-               ns_clientmgr_destroy(&mgr->clientmgrs[i]);
+               ns_clientmgr_detach(&mgr->clientmgrs[i]);
        }
        isc_mem_put(mgr->mctx, mgr->clientmgrs,
                    mgr->ncpus * sizeof(mgr->clientmgrs[0]));
@@ -451,6 +451,10 @@ ns_interfacemgr_shutdown(ns_interfacemgr_t *mgr) {
        if (mgr->route != NULL) {
                isc_nm_cancelread(mgr->route);
        }
+
+       for (size_t i = 0; i < mgr->ncpus; i++) {
+               ns_clientmgr_shutdown(mgr->clientmgrs[i]);
+       }
 }
 
 static void