LU-9120 lnet: set health value from user space

Add commands to lnetctl to set the health value.

for local NIs:
 lnetctl net set --nid <nid> --health <value>

for peer NIs:
 lnetctl peer set --nid <nid> --health <value>

Test-Parameters: forbuildonly
Signed-off-by: Amir Shehata <ashehata@whamcloud.com>
Change-Id: I06e1238df54c94bcfecadd84fbaa30cc1ce4dd68
Reviewed-on: https://review.whamcloud.com/32863
Tested-by: Jenkins
Reviewed-by: Sonia Sharma <sharmaso@whamcloud.com>
Reviewed-by: Olaf Weber <olaf.weber@hpe.com>

diff --git a/lnet/lnet/api-ni.c b/lnet/lnet/api-ni.c
index a0fe3a5..e0b2a22 100644 (file)
--- a/lnet/lnet/api-ni.c
+++ b/lnet/lnet/api-ni.c
@@ -3598,6 +3598,9 @@ LNetCtl(unsigned int cmd, void *arg)
                        value = LNET_MAX_HEALTH_VALUE;
                else
                        value = cfg->rh_value;
+               CDEBUG(D_NET, "Manually setting healthv to %d for %s:%s. all = %d\n",
+                      value, (cfg->rh_type == LNET_HEALTH_TYPE_LOCAL_NI) ?
+                      "local" : "peer", libcfs_nid2str(cfg->rh_nid), cfg->rh_all);
                mutex_lock(&the_lnet.ln_api_mutex);
                if (cfg->rh_type == LNET_HEALTH_TYPE_LOCAL_NI)
                        lnet_ni_set_healthv(cfg->rh_nid, value,

diff --git a/lnet/lnet/peer.c b/lnet/lnet/peer.c
index 0159101..6d32f19 100644 (file)
--- a/lnet/lnet/peer.c
+++ b/lnet/lnet/peer.c
@@ -3495,6 +3495,10 @@ lnet_peer_ni_set_healthv(lnet_nid_t nid, int value, bool all)
        if (!all) {
                lnet_net_lock(LNET_LOCK_EX);
                lpni = lnet_find_peer_ni_locked(nid);
+               if (!lpni) {
+                       lnet_net_unlock(LNET_LOCK_EX);
+                       return;
+               }
                atomic_set(&lpni->lpni_healthv, value);
                lnet_peer_ni_add_to_recoveryq_locked(lpni);
                lnet_peer_ni_decref_locked(lpni);

diff --git a/lnet/utils/lnetconfig/liblnetconfig.c b/lnet/utils/lnetconfig/liblnetconfig.c
index c4a3cdc..e481cf7 100644 (file)
--- a/lnet/utils/lnetconfig/liblnetconfig.c
+++ b/lnet/utils/lnetconfig/liblnetconfig.c
@@ -2028,6 +2028,61 @@ out:
        return rc;
 }
 
+static int
+lustre_lnet_config_healthv(int value, bool all, lnet_nid_t nid,
+                          enum lnet_health_type type, char *name,
+                          int seq_no, struct cYAML **err_rc)
+{
+       struct lnet_ioctl_reset_health_cfg data;
+       int rc = LUSTRE_CFG_RC_NO_ERR;
+       char err_str[LNET_MAX_STR_LEN];
+
+       snprintf(err_str, sizeof(err_str), "\"success\"");
+
+       LIBCFS_IOC_INIT_V2(data, rh_hdr);
+       data.rh_type = type;
+       data.rh_all = all;
+       data.rh_value = value;
+       data.rh_nid = nid;
+
+       rc = l_ioctl(LNET_DEV_ID, IOC_LIBCFS_SET_HEALHV, &data);
+       if (rc != 0) {
+               rc = -errno;
+               snprintf(err_str,
+                        sizeof(err_str), "Can not configure health value");
+       }
+
+       cYAML_build_error(rc, seq_no, ADD_CMD, name, err_str, err_rc);
+
+       return rc;
+}
+
+int lustre_lnet_config_ni_healthv(int value, bool all, char *ni_nid, int seq_no,
+                                 struct cYAML **err_rc)
+{
+       lnet_nid_t nid;
+       if (ni_nid)
+               nid = libcfs_str2nid(ni_nid);
+       else
+               nid = LNET_NID_ANY;
+       return lustre_lnet_config_healthv(value, all, nid,
+                                         LNET_HEALTH_TYPE_LOCAL_NI,
+                                         "ni healthv", seq_no, err_rc);
+}
+
+int lustre_lnet_config_peer_ni_healthv(int value, bool all, char *lpni_nid,
+                                      int seq_no, struct cYAML **err_rc)
+{
+       lnet_nid_t nid;
+       if (lpni_nid)
+               nid = libcfs_str2nid(lpni_nid);
+       else
+               nid = LNET_NID_ANY;
+       return lustre_lnet_config_healthv(value, all, nid,
+                                         LNET_HEALTH_TYPE_PEER_NI,
+                                         "peer_ni healthv", seq_no, err_rc);
+}
+
 static bool
 add_msg_stats_to_yaml_blk(struct cYAML *yaml,
                          struct lnet_ioctl_comm_count *counts)

diff --git a/lnet/utils/lnetconfig/liblnetconfig.h b/lnet/utils/lnetconfig/liblnetconfig.h
index 6d67302..8ec4069 100644 (file)
--- a/lnet/utils/lnetconfig/liblnetconfig.h
+++ b/lnet/utils/lnetconfig/liblnetconfig.h
@@ -225,6 +225,36 @@ int lustre_lnet_show_numa_range(int seq_no, struct cYAML **show_rc,
                                struct cYAML **err_rc);
 
 /*
+ * lustre_lnet_config_ni_healthv
+ *   set the health value of the NI. -1 resets the value to maximum.
+ *
+ *   value: health value to set.
+ *   all: true to set all local NIs to that value.
+ *   ni_nid: NI NID to set its health value. all parameter always takes
+ *   precedence
+ *   seq_no - sequence number of the request
+ *   err_rc - [OUT] struct cYAML tree describing the error. Freed by
+ *   caller
+ */
+int lustre_lnet_config_ni_healthv(int value, bool all, char *ni_nid,
+                                 int seq_no, struct cYAML **err_rc);
+
+/*
+ * lustre_lnet_config_peer_ni_healthv
+ *   set the health value of the peer NI. -1 resets the value to maximum.
+ *
+ *   value: health value to set.
+ *   all: true to set all local NIs to that value.
+ *   pni_nid: Peer NI NID to set its health value. all parameter always takes
+ *   precedence
+ *   seq_no - sequence number of the request
+ *   err_rc - [OUT] struct cYAML tree describing the error. Freed by
+ *   caller
+ */
+int lustre_lnet_config_peer_ni_healthv(int value, bool all, char *pni_nid,
+                                      int seq_no, struct cYAML **err_rc);
+
+/*
  * lustre_lnet_config_hsensitivity
  *   sets the health sensitivity; the value by which to decrement the
  *   health value of a local or peer NI. If 0 then health is turned off

diff --git a/lnet/utils/lnetctl.c b/lnet/utils/lnetctl.c
index 7633082..e6e4255 100644 (file)
--- a/lnet/utils/lnetctl.c
+++ b/lnet/utils/lnetctl.c
@@ -75,7 +75,8 @@ static int jt_set(int argc, char **argv);
 static int jt_stats(int argc, char **argv);
 static int jt_global(int argc, char **argv);
 static int jt_peers(int argc, char **argv);
-
+static int jt_set_ni_value(int argc, char **argv);
+static int jt_set_peer_ni_value(int argc, char **argv);
 
 command_t cmd_list[] = {
        {"lnet", jt_lnet, 0, "lnet {configure | unconfigure} [--all]"},
@@ -141,6 +142,10 @@ command_t net_cmds[] = {
         "\t--net: net name (e.g. tcp0) to filter on\n"
         "\t--verbose: display detailed output per network."
                       " Optional argument of '2' outputs more stats\n"},
+       {"set", jt_set_ni_value, 0, "set local NI specific parameter\n"
+        "\t--nid: NI NID to set the\n"
+        "\t--health: specify health value to set\n"
+        "\t--all: set all NIs value to the one specified\n"},
        { 0, 0, 0, NULL }
 };
 
@@ -206,6 +211,10 @@ command_t peer_cmds[] = {
         "\t--verbose: display detailed output per peer."
                       " Optional argument of '2' outputs more stats\n"},
        {"list", jt_list_peer, 0, "list all peers\n"},
+       {"set", jt_set_peer_ni_value, 0, "set peer ni specific parameter\n"
+        "\t--nid: Peer NI NID to set the\n"
+        "\t--health: specify health value to set\n"
+        "\t--all: set all peer_nis values to the one specified\n"},
        { 0, 0, 0, NULL }
 };
 
@@ -953,6 +962,63 @@ static int jt_show_route(int argc, char **argv)
        return rc;
 }
 
+static int set_value_helper(int argc, char **argv,
+                           int (*cb)(int, bool, char*, int, struct cYAML**))
+{
+       char *nid = NULL;
+       long int healthv = -1;
+       bool all = false;
+       int rc, opt;
+       struct cYAML *err_rc = NULL;
+
+       const char *const short_options = "h:n:a";
+       static const struct option long_options[] = {
+               { .name = "nid", .has_arg = required_argument, .val = 'n' },
+               { .name = "health", .has_arg = required_argument, .val = 'h' },
+               { .name = "all", .has_arg = no_argument, .val = 'a' },
+               { .name = NULL } };
+
+       rc = check_cmd(net_cmds, "net", "set", 0, argc, argv);
+       if (rc)
+               return rc;
+
+       while ((opt = getopt_long(argc, argv, short_options,
+                                  long_options, NULL)) != -1) {
+               switch (opt) {
+               case 'n':
+                       nid = optarg;
+                       break;
+               case 'h':
+                       if (parse_long(argv[optind++], &healthv) != 0)
+                               healthv = -1;
+                       break;
+               case 'a':
+                       all = true;
+               default:
+                       return 0;
+               }
+       }
+
+       rc = cb(healthv, all, nid, -1, &err_rc);
+
+       if (rc != LUSTRE_CFG_RC_NO_ERR)
+               cYAML_print_tree2file(stderr, err_rc);
+
+       cYAML_free_tree(err_rc);
+
+       return rc;
+}
+
+static int jt_set_ni_value(int argc, char **argv)
+{
+       return set_value_helper(argc, argv, lustre_lnet_config_ni_healthv);
+}
+
+static int jt_set_peer_ni_value(int argc, char **argv)
+{
+       return set_value_helper(argc, argv, lustre_lnet_config_peer_ni_healthv);
+}
+
 static int jt_show_net(int argc, char **argv)
 {
        char *network = NULL;