X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lnet%2Futils%2Flnetctl.c;h=504e189dd5d52e2a4f06f6ac265c5e11f02d7a88;hb=b7f8d156db696fcc15fd37cfdfbee6549148fb69;hp=a385f4000fdfaf86006815b204aa0c14a71e1682;hpb=ccabce23bd9e366c345c852f565766a799f61238;p=fs%2Flustre-release.git diff --git a/lnet/utils/lnetctl.c b/lnet/utils/lnetctl.c index a385f40..504e189 100644 --- a/lnet/utils/lnetctl.c +++ b/lnet/utils/lnetctl.c @@ -48,11 +48,16 @@ static int jt_show_net(int argc, char **argv); static int jt_show_routing(int argc, char **argv); static int jt_show_stats(int argc, char **argv); static int jt_show_peer(int argc, char **argv); +static int jt_show_recovery(int argc, char **argv); static int jt_show_global(int argc, char **argv); static int jt_set_tiny(int argc, char **argv); static int jt_set_small(int argc, char **argv); static int jt_set_large(int argc, char **argv); static int jt_set_numa(int argc, char **argv); +static int jt_set_retry_count(int argc, char **argv); +static int jt_set_transaction_to(int argc, char **argv); +static int jt_set_recov_intrv(int argc, char **argv); +static int jt_set_hsensitivity(int argc, char **argv); static int jt_add_peer_nid(int argc, char **argv); static int jt_del_peer_nid(int argc, char **argv); static int jt_set_max_intf(int argc, char **argv); @@ -69,10 +74,12 @@ static int jt_route(int argc, char **argv); static int jt_net(int argc, char **argv); static int jt_routing(int argc, char **argv); static int jt_set(int argc, char **argv); +static int jt_debug(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]"}, @@ -85,6 +92,7 @@ command_t cmd_list[] = { {"import", jt_import, 0, "import FILE.yaml"}, {"export", jt_export, 0, "export FILE.yaml"}, {"stats", jt_stats, 0, "stats {show | help}"}, + {"debug", jt_debug, 0, "debug recovery {local | peer}"}, {"global", jt_global, 0, "global {show | help}"}, {"peer", jt_peers, 0, "peer {add | del | show | help}"}, {"ping", jt_ping, 0, "ping nid,[nid,...]"}, @@ -138,6 +146,10 @@ 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 } }; @@ -151,6 +163,13 @@ command_t stats_cmds[] = { { 0, 0, 0, NULL } }; +command_t debug_cmds[] = { + {"recovery", jt_show_recovery, 0, "list recovery queues\n" + "\t--local : list local recovery queue\n" + "\t--peer : list peer recovery queue\n"}, + { 0, 0, 0, NULL } +}; + command_t global_cmds[] = { {"show", jt_show_global, 0, "show global variables\n"}, { 0, 0, 0, NULL } @@ -174,6 +193,16 @@ command_t set_cmds[] = { {"discovery", jt_set_discovery, 0, "enable/disable peer discovery\n" "\t0 - disable peer discovery\n" "\t1 - enable peer discovery (default)\n"}, + {"retry_count", jt_set_retry_count, 0, "number of retries\n" + "\t0 - turn of retries\n" + "\t>0 - number of retries\n"}, + {"transaction_timeout", jt_set_transaction_to, 0, "Message/Response timeout\n" + "\t>0 - timeout in seconds\n"}, + {"health_sensitivity", jt_set_hsensitivity, 0, "sensitivity to failure\n" + "\t0 - turn off health evaluation\n" + "\t>0 - sensitivity value not more than 1000\n"}, + {"recovery_interval", jt_set_recov_intrv, 0, "interval to ping in seconds (at least 1)\n" + "\t>0 - time in seconds between pings\n"}, { 0, 0, 0, NULL } }; @@ -183,16 +212,22 @@ command_t peer_cmds[] = { "\t NID in the list becomes the Primary NID of a newly created\n" "\t peer. \n" "\t--nid: one or more peer NIDs\n" - "\t--non_mr: create this peer as not Multi-Rail capable\n"}, + "\t--non_mr: create this peer as not Multi-Rail capable\n" + "\t--ip2nets: specify a range of nids per peer"}, {"del", jt_del_peer_nid, 0, "delete a peer NID\n" "\t--prim_nid: Primary NID of the peer.\n" "\t--nid: list of NIDs to remove. If none provided,\n" - "\t peer is deleted\n"}, + "\t peer is deleted\n" + "\t--ip2nets: specify a range of nids per peer"}, {"show", jt_show_peer, 0, "show peer information\n" "\t--nid: NID of peer to filter on.\n" "\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 } }; @@ -326,6 +361,118 @@ static int jt_set_numa(int argc, char **argv) return rc; } +static int jt_set_recov_intrv(int argc, char **argv) +{ + long int value; + int rc; + struct cYAML *err_rc = NULL; + + rc = check_cmd(set_cmds, "set", "recovery_interval", 2, argc, argv); + if (rc) + return rc; + + rc = parse_long(argv[1], &value); + if (rc != 0) { + cYAML_build_error(-1, -1, "parser", "set", + "cannot parse recovery interval value", &err_rc); + cYAML_print_tree2file(stderr, err_rc); + cYAML_free_tree(err_rc); + return -1; + } + + rc = lustre_lnet_config_recov_intrv(value, -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_hsensitivity(int argc, char **argv) +{ + long int value; + int rc; + struct cYAML *err_rc = NULL; + + rc = check_cmd(set_cmds, "set", "health_sensitivity", 2, argc, argv); + if (rc) + return rc; + + rc = parse_long(argv[1], &value); + if (rc != 0) { + cYAML_build_error(-1, -1, "parser", "set", + "cannot parse health sensitivity value", &err_rc); + cYAML_print_tree2file(stderr, err_rc); + cYAML_free_tree(err_rc); + return -1; + } + + rc = lustre_lnet_config_hsensitivity(value, -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_transaction_to(int argc, char **argv) +{ + long int value; + int rc; + struct cYAML *err_rc = NULL; + + rc = check_cmd(set_cmds, "set", "transaction_timeout", 2, argc, argv); + if (rc) + return rc; + + rc = parse_long(argv[1], &value); + if (rc != 0) { + cYAML_build_error(-1, -1, "parser", "set", + "cannot parse transaction timeout value", &err_rc); + cYAML_print_tree2file(stderr, err_rc); + cYAML_free_tree(err_rc); + return -1; + } + + rc = lustre_lnet_config_transaction_to(value, -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_retry_count(int argc, char **argv) +{ + long int value; + int rc; + struct cYAML *err_rc = NULL; + + rc = check_cmd(set_cmds, "set", "retry_count", 2, argc, argv); + if (rc) + return rc; + + rc = parse_long(argv[1], &value); + if (rc != 0) { + cYAML_build_error(-1, -1, "parser", "set", + "cannot parse retry_count value", &err_rc); + cYAML_print_tree2file(stderr, err_rc); + cYAML_free_tree(err_rc); + return -1; + } + + rc = lustre_lnet_config_retry_count(value, -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_discovery(int argc, char **argv) { long int value; @@ -569,6 +716,8 @@ static int jt_add_route(int argc, char **argv) continue; } break; + case '?': + print_help(route_cmds, "route", "add"); default: return 0; } @@ -670,6 +819,8 @@ static int jt_add_ni(int argc, char **argv) strlen(optarg), 0, UINT_MAX, &global_cpts); break; + case '?': + print_help(net_cmds, "net", "add"); default: return 0; } @@ -725,6 +876,8 @@ static int jt_del_route(int argc, char **argv) case 'g': gateway = optarg; break; + case '?': + print_help(route_cmds, "route", "del"); default: return 0; } @@ -773,6 +926,8 @@ static int jt_del_ni(int argc, char **argv) goto out; } break; + case '?': + print_help(net_cmds, "net", "del"); default: return 0; } @@ -837,13 +992,113 @@ static int jt_show_route(int argc, char **argv) case 'v': detail = 1; break; + case '?': + print_help(route_cmds, "route", "show"); default: return 0; } } rc = lustre_lnet_show_route(network, gateway, hop, prio, detail, -1, - &show_rc, &err_rc); + &show_rc, &err_rc, false); + + if (rc != LUSTRE_CFG_RC_NO_ERR) + cYAML_print_tree2file(stderr, err_rc); + else if (show_rc) + cYAML_print_tree(show_rc); + + cYAML_free_tree(err_rc); + cYAML_free_tree(show_rc); + + 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 = "t:n:a"; + static const struct option long_options[] = { + { .name = "nid", .has_arg = required_argument, .val = 'n' }, + { .name = "health", .has_arg = required_argument, .val = 't' }, + { .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 't': + if (parse_long(optarg, &healthv) != 0) + healthv = -1; + break; + case 'a': + all = true; + break; + 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_recovery(int argc, char **argv) +{ + int rc, opt; + struct cYAML *err_rc = NULL, *show_rc = NULL; + + const char *const short_options = "lp"; + static const struct option long_options[] = { + { .name = "local", .has_arg = no_argument, .val = 'l' }, + { .name = "peer", .has_arg = no_argument, .val = 'p' }, + { .name = NULL } }; + + rc = check_cmd(debug_cmds, "recovery", NULL, 0, argc, argv); + if (rc) + return rc; + + while ((opt = getopt_long(argc, argv, short_options, + long_options, NULL)) != -1) { + switch (opt) { + case 'l': + rc = lustre_lnet_show_local_ni_recovq(-1, &show_rc, &err_rc); + break; + case 'p': + rc = lustre_lnet_show_peer_ni_recovq(-1, &show_rc, &err_rc); + break; + default: + return 0; + } + } if (rc != LUSTRE_CFG_RC_NO_ERR) cYAML_print_tree2file(stderr, err_rc); @@ -888,12 +1143,15 @@ static int jt_show_net(int argc, char **argv) detail = 1; } break; + case '?': + print_help(net_cmds, "net", "show"); default: return 0; } } - rc = lustre_lnet_show_net(network, (int) detail, -1, &show_rc, &err_rc); + rc = lustre_lnet_show_net(network, (int) detail, -1, &show_rc, &err_rc, + false); if (rc != LUSTRE_CFG_RC_NO_ERR) cYAML_print_tree2file(stderr, err_rc); @@ -915,7 +1173,7 @@ static int jt_show_routing(int argc, char **argv) if (rc) return rc; - rc = lustre_lnet_show_routing(-1, &show_rc, &err_rc); + rc = lustre_lnet_show_routing(-1, &show_rc, &err_rc, false); if (rc != LUSTRE_CFG_RC_NO_ERR) cYAML_print_tree2file(stderr, err_rc); @@ -977,6 +1235,30 @@ static int jt_show_global(int argc, char **argv) goto out; } + rc = lustre_lnet_show_retry_count(-1, &show_rc, &err_rc); + if (rc != LUSTRE_CFG_RC_NO_ERR) { + cYAML_print_tree2file(stderr, err_rc); + goto out; + } + + rc = lustre_lnet_show_transaction_to(-1, &show_rc, &err_rc); + if (rc != LUSTRE_CFG_RC_NO_ERR) { + cYAML_print_tree2file(stderr, err_rc); + goto out; + } + + rc = lustre_lnet_show_hsensitivity(-1, &show_rc, &err_rc); + if (rc != LUSTRE_CFG_RC_NO_ERR) { + cYAML_print_tree2file(stderr, err_rc); + goto out; + } + + rc = lustre_lnet_show_recov_intrv(-1, &show_rc, &err_rc); + if (rc != LUSTRE_CFG_RC_NO_ERR) { + cYAML_print_tree2file(stderr, err_rc); + goto out; + } + if (show_rc) cYAML_print_tree(show_rc); @@ -1042,6 +1324,17 @@ static int jt_stats(int argc, char **argv) return Parser_execarg(argc - 1, &argv[1], stats_cmds); } +static int jt_debug(int argc, char **argv) +{ + int rc; + + rc = check_cmd(debug_cmds, "recovery", NULL, 2, argc, argv); + if (rc) + return rc; + + return Parser_execarg(argc - 1, &argv[1], debug_cmds); +} + static int jt_global(int argc, char **argv) { int rc; @@ -1165,32 +1458,44 @@ static int jt_export(int argc, char **argv) int rc; FILE *f = NULL; int opt; + bool backup = false; + char *file = NULL; - const char *const short_options = "h"; + const char *const short_options = "bh"; static const struct option long_options[] = { + { .name = "backup", .has_arg = no_argument, .val = 'b' }, { .name = "help", .has_arg = no_argument, .val = 'h' }, { .name = NULL } }; while ((opt = getopt_long(argc, argv, short_options, long_options, NULL)) != -1) { switch (opt) { + case 'b': + backup = true; + break; case 'h': + default: printf("export > FILE.yaml : export configuration\n" + "\t--backup: export only what's necessary for reconfig\n" "\t--help: display this help\n"); return 0; - default: - return 0; } } - if (argc >= 2) { - f = fopen(argv[1], "w"); + if (backup && argc >= 3) + file = argv[2]; + else if (!backup && argc >= 2) + file = argv[1]; + else + f = stdout; + + if (file) { + f = fopen(file, "w"); if (f == NULL) return -1; - } else - f = stdout; + } - rc = lustre_lnet_show_net(NULL, 2, -1, &show_rc, &err_rc); + rc = lustre_lnet_show_net(NULL, 2, -1, &show_rc, &err_rc, backup); if (rc != LUSTRE_CFG_RC_NO_ERR) { cYAML_print_tree2file(stderr, err_rc); cYAML_free_tree(err_rc); @@ -1198,21 +1503,21 @@ static int jt_export(int argc, char **argv) } rc = lustre_lnet_show_route(NULL, NULL, -1, -1, 1, -1, &show_rc, - &err_rc); + &err_rc, backup); if (rc != LUSTRE_CFG_RC_NO_ERR) { cYAML_print_tree2file(stderr, err_rc); cYAML_free_tree(err_rc); err_rc = NULL; } - rc = lustre_lnet_show_routing(-1, &show_rc, &err_rc); + rc = lustre_lnet_show_routing(-1, &show_rc, &err_rc, backup); if (rc != LUSTRE_CFG_RC_NO_ERR) { cYAML_print_tree2file(stderr, err_rc); cYAML_free_tree(err_rc); err_rc = NULL; } - rc = lustre_lnet_show_peer(NULL, 2, -1, &show_rc, &err_rc); + rc = lustre_lnet_show_peer(NULL, 2, -1, &show_rc, &err_rc, backup); if (rc != LUSTRE_CFG_RC_NO_ERR) { cYAML_print_tree2file(stderr, err_rc); cYAML_free_tree(err_rc); @@ -1259,12 +1564,14 @@ static int jt_add_peer_nid(int argc, char **argv) struct cYAML *err_rc = NULL; int rc = LUSTRE_CFG_RC_NO_ERR, opt, i; bool non_mr = false; + bool ip2nets = false, nid_list = false, prim_nid_present = false; const char *const short_opts = "k:mn:"; const struct option long_opts[] = { { .name = "prim_nid", .has_arg = required_argument, .val = 'k' }, { .name = "non_mr", .has_arg = no_argument, .val = 'm' }, { .name = "nid", .has_arg = required_argument, .val = 'n' }, + { .name = "ip2nets", .has_arg = required_argument, .val = 'i' }, { .name = NULL } }; rc = check_cmd(peer_cmds, "peer", "add", 2, argc, argv); @@ -1275,9 +1582,32 @@ static int jt_add_peer_nid(int argc, char **argv) long_opts, NULL)) != -1) { switch (opt) { case 'k': + prim_nid_present = true; + if (ip2nets) { + cYAML_build_error(-1, -1, "peer", "add", + "ip2nets can not be specified" + " along side prim_nid parameter.", + &err_rc); + goto failed; + } prim_nid = optarg; break; + case 'i': case 'n': + if (opt == 'i') + ip2nets = true; + + if (opt == 'n') + nid_list = true; + + if (ip2nets && (nid_list || prim_nid_present)) { + cYAML_build_error(-1, -1, "peer", "add", + "ip2nets can not be specified" + " along side nid or prim_nid" + " parameters", &err_rc); + goto failed; + } + size = lustre_lnet_parse_nids(optarg, nids, size, &nids2); if (nids2 == NULL) @@ -1288,13 +1618,23 @@ static int jt_add_peer_nid(int argc, char **argv) case 'm': non_mr = true; break; + case '?': + print_help(peer_cmds, "peer", "add"); default: return 0; } } + for (; optind < argc; optind++) { + size = lustre_lnet_parse_nids(argv[optind], nids, size, + &nids2); + if (nids2 == NULL) + goto failed; + nids = nids2; + } + rc = lustre_lnet_config_peer_nid(prim_nid, nids, size, - !non_mr, -1, &err_rc); + !non_mr, ip2nets, -1, &err_rc); failed: if (nids) { @@ -1318,11 +1658,13 @@ static int jt_del_peer_nid(int argc, char **argv) char **nids = NULL, **nids2 = NULL; struct cYAML *err_rc = NULL; int rc = LUSTRE_CFG_RC_NO_ERR, opt, i, size = 0; + bool ip2nets = false, nid_list = false, prim_nid_present = false; const char *const short_opts = "k:n:"; const struct option long_opts[] = { { .name = "prim_nid", .has_arg = required_argument, .val = 'k' }, { .name = "nid", .has_arg = required_argument, .val = 'n' }, + { .name = "ip2nets", .has_arg = required_argument, .val = 'i' }, { .name = NULL } }; rc = check_cmd(peer_cmds, "peer", "del", 2, argc, argv); @@ -1333,9 +1675,31 @@ static int jt_del_peer_nid(int argc, char **argv) long_opts, NULL)) != -1) { switch (opt) { case 'k': + prim_nid_present = true; + if (ip2nets) { + cYAML_build_error(-1, -1, "peer", "add", + "ip2nets can not be specified" + " along side prim_nid parameter.", + &err_rc); + goto failed; + } prim_nid = optarg; break; + case 'i': case 'n': + if (opt == 'i') + ip2nets = true; + + if (opt == 'n') + nid_list = true; + + if (ip2nets && (nid_list || prim_nid_present)) { + cYAML_build_error(-1, -1, "peer", "add", + "ip2nets can not be specified" + " along side nid or prim_nid" + " parameters", &err_rc); + goto failed; + } size = lustre_lnet_parse_nids(optarg, nids, size, &nids2); if (nids2 == NULL) @@ -1343,12 +1707,22 @@ static int jt_del_peer_nid(int argc, char **argv) nids = nids2; rc = LUSTRE_CFG_RC_OUT_OF_MEM; break; + case '?': + print_help(peer_cmds, "peer", "del"); default: return 0; } } - rc = lustre_lnet_del_peer_nid(prim_nid, nids, size, -1, &err_rc); + for (; optind < argc; optind++) { + size = lustre_lnet_parse_nids(argv[optind], nids, size, + &nids2); + if (nids2 == NULL) + goto failed; + nids = nids2; + } + + rc = lustre_lnet_del_peer_nid(prim_nid, nids, size, ip2nets, -1, &err_rc); failed: if (nids) { @@ -1398,12 +1772,15 @@ static int jt_show_peer(int argc, char **argv) detail = 1; } break; + case '?': + print_help(peer_cmds, "peer", "show"); default: return 0; } } - rc = lustre_lnet_show_peer(nid, (int) detail, -1, &show_rc, &err_rc); + rc = lustre_lnet_show_peer(nid, (int) detail, -1, &show_rc, &err_rc, + false); if (rc != LUSTRE_CFG_RC_NO_ERR) cYAML_print_tree2file(stderr, err_rc);