July 2012 - cluster-commits - Fedora Mailing-Lists

cluster: STABLE32 - qdiskd: allow master to failover quickly when using master_wins

by Fabio M. Di Nitto

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=e8af462b753... Commit: e8af462b7531f87b5cb20f7204eeb4b520591da9 Parent: 03e2215bd277fd79b8a6ee70a49de711e0f343ad Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 10:27:57 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 10:27:57 2012 +0200 qdiskd: allow master to failover quickly when using master_wins in case of master_wins and we are shutting down the master qdiskd, there is a small window in which the other node is not quorate because qdiskd has not become master yet. this patch allows the master qdiskd to communicate to the other nodes that it is going away and gives enough time to elect a new master before dieing. the process itself is safe and the worst case scenario the cluster will behave as-is now (temporary loss of quorum), otherwise a fast switch will take place. Resolves: rhbz#814807 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> Reviewed-by: Lon Hohberger <lhh(a)redhat.com> --- cman/qdisk/disk.h | 5 +++- cman/qdisk/main.c | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 75 insertions(+), 2 deletions(-) diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index fd80fa6..6bed41d 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -24,9 +24,12 @@ typedef enum { S_INIT = 0x2, // Initializing. Hold your fire. /* vvv Fencing will kill a node */ S_RUN = 0x5, // I think I'm running. - S_MASTER= 0x6 // I know I'm running, and have advertised to + S_MASTER= 0x6, // I know I'm running, and have advertised to // CMAN the availability of the disk vote for my // partition. + S_EXIT = 0x7 // trigger master re-election before exit + // status is set only by master in master-win | auto-masterwin + // and next status _must_ be S_NONE } disk_node_state_t; diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index 32677a2..16c26e4 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -196,7 +196,8 @@ read_node_blocks(qd_ctx *ctx, node_info_t *ni, int max) continue; /* Unchanged timestamp: miss */ - if (sb->ps_timestamp == ni[x].ni_last_seen) { + if ((sb->ps_timestamp == ni[x].ni_last_seen) && + (ni[x].ni_state != S_EXIT)) { /* XXX check for average + allow grace */ ni[x].ni_misses++; if (ni[x].ni_misses > 1) { @@ -231,6 +232,22 @@ check_transitions(qd_ctx *ctx, node_info_t *ni, int max, memb_mask_t mask) for (x = 0; x < max; x++) { /* + Case 0: check if master node is about to leave + */ + if (ni[x].ni_state == S_EXIT) { + logt_print(LOG_NOTICE, "Node %d is about to leave\n", ni[x].ni_status.ps_nodeid); + ni[x].ni_evil_incarnation = 0; + ni[x].ni_incarnation = 0; + ni[x].ni_seen = 0; + ni[x].ni_misses = 0; + ni[x].ni_state = S_NONE; + if (mask) + clear_bit(mask, (ni[x].ni_status.ps_nodeid-1), + sizeof(memb_mask_t)); + continue; + } + + /* Case 1: check to see if the node is still up according to our internal state, but has been evicted by the master or cleanly shut down @@ -1269,6 +1286,50 @@ quorum_loop(qd_ctx *ctx, node_info_t *ni, int max) /** + Tell the other nodes to elect a new master != me. + */ +static int +quorum_reelect_master(qd_ctx *ctx, node_info_t *ni, int max) +{ + if (qd_write_status(ctx, ctx->qc_my_id, S_EXIT, + NULL, NULL, NULL) != 0) { + logt_print(LOG_WARNING, + "Error writing to quorum disk during reelect_master\n"); + } + + while (1) { + int master, x; + int found = 0; + int low_id, count; + + read_node_blocks(ctx, ni, max); + + for (x = 0; x < max; x++) { + if (ni[x].ni_state >= S_RUN) { + found = 1; + } + } + + if (!found) { + logt_print(LOG_DEBUG, "No other nodes are active. Exiting\n"); + break; + } + + master = master_exists(ctx, ni, max, &low_id, &count); + if (master) { + logt_print(LOG_DEBUG, "New master elected: %d\n", master); + break; + } + /* + * give time for message to be read + */ + sleep(1); + } + + return 0; +} + +/** Tell the other nodes we're done (safely!). */ static int @@ -2173,6 +2234,15 @@ main(int argc, char **argv) io_nanny_start(ch_user, ctx.qc_tko * ctx.qc_interval); if (quorum_loop(&ctx, ni, MAX_NODES_DISK) == 0) { + /* + * if we are master and we are in master-win mode, + * request other qdiskd to elect a new one + */ + if ((ctx.qc_status == S_MASTER) && + ((ctx.qc_flags & RF_MASTER_WINS) || + (ctx.qc_flags & RF_AUTO_MASTER_WINS))) { + quorum_reelect_master(&ctx, ni, MAX_NODES_DISK); + } /* Only clean up if we're exiting w/o error) */ logt_print(LOG_NOTICE, "Unregistering quorum device.\n"); cman_unregister_quorum_device(ctx.qc_cman_admin);

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - cman init: allow dlm hash table sizes to be tunable at startup

by Fabio M. Di Nitto

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=332db7754d6... Commit: 332db7754d6210991c34be056789aa1b33b26a0a Parent: 77bb92ad9dbad9c8755e45211214ab1e099cdb0e Author: Fabio M. Di Nitto <fdinitto(a)redhat.com> AuthorDate: Tue Jul 24 08:55:27 2012 +0200 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Tue Jul 24 08:55:27 2012 +0200 cman init: allow dlm hash table sizes to be tunable at startup Resolves: rhbz#836963 Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/init.d/cman | 43 ++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 42 insertions(+), 1 deletions(-) diff --git a/cman/init.d/cman b/cman/init.d/cman index 4727d22..2101eb7 100755 --- a/cman/init.d/cman +++ b/cman/init.d/cman @@ -297,6 +297,37 @@ fence_join_enabled() fi } +tune_dlm_hash_sizes() +{ + dlmdir="/sys/kernel/config/dlm/cluster" + maxloop=21 + + if [ -z "$DLM_LKBTBL_SIZE" ] && \ + [ -z "$DLM_RSBTBL_SIZE" ] && \ + [ -z "$DLM_DIRTBL_SIZE" ]; then + return 0 + fi + for i in $(seq 1 $maxloop); do + if [ -d $dlmdir ]; then + break + fi + sleep 0.5 + done + if [ "$i" = "$maxloop" ]; then + return 1 + fi + if [ -n "$DLM_LKBTBL_SIZE" ]; then + echo $DLM_LKBTBL_SIZE > $dlmdir/lkbtbl_size + fi + if [ -n "$DLM_RSBTBL_SIZE" ]; then + echo $DLM_RSBTBL_SIZE > $dlmdir/rsbtbl_size + fi + if [ -n "$DLM_DIRTBL_SIZE" ]; then + echo $DLM_DIRTBL_SIZE > $dlmdir/dirtbl_size + fi + return 0 +} + start() { echo "Starting cluster: " @@ -391,7 +422,17 @@ start() return 1 fi fi - + + echo -n " Tuning DLM... " + tune_dlm_hash_sizes + if [ $? -eq 0 ] + then + echo "done" + else + echo "failed" + return 1 + fi + return 0 }

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - fenced: fix double free when second method has no devices

by David Teigland

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=77bb92ad9db... Commit: 77bb92ad9dbad9c8755e45211214ab1e099cdb0e Parent: 6c9717cbbca38ee94911cbbc481c310f6375d560 Author: David Teigland <teigland(a)redhat.com> AuthorDate: Mon Jul 23 17:01:57 2012 -0500 Committer: David Teigland <teigland(a)redhat.com> CommitterDate: Mon Jul 23 17:01:57 2012 -0500 fenced: fix double free when second method has no devices When the second fence method is empty, the already freed device is freed again, causing a segfault. bz 809390 Signed-off-by: David Teigland <teigland(a)redhat.com> --- fence/fenced/agent.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/fence/fenced/agent.c b/fence/fenced/agent.c index 753b915..de4467c 100644 --- a/fence/fenced/agent.c +++ b/fence/fenced/agent.c @@ -361,8 +361,10 @@ int dispatch_fence_agent(char *victim, int force) device = NULL; } - if (device) + if (device) { free(device); + device = NULL; + } free(method); if (!error) {

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - fenced: increase MAX_DEVICES to 8

by David Teigland

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=6c9717cbbca... Commit: 6c9717cbbca38ee94911cbbc481c310f6375d560 Parent: b5253b0c74175ecda233f573fb0afa733b60eda2 Author: David Teigland <teigland(a)redhat.com> AuthorDate: Mon Jul 23 10:23:56 2012 -0500 Committer: David Teigland <teigland(a)redhat.com> CommitterDate: Mon Jul 23 16:36:44 2012 -0500 fenced: increase MAX_DEVICES to 8 Make this match rhel6. bz 821857 Signed-off-by: David Teigland <teigland(a)redhat.com> --- fence/fenced/agent.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/fence/fenced/agent.c b/fence/fenced/agent.c index e3aab9a..753b915 100644 --- a/fence/fenced/agent.c +++ b/fence/fenced/agent.c @@ -27,7 +27,7 @@ #include "ccs.h" #define MAX_METHODS 8 -#define MAX_DEVICES 4 +#define MAX_DEVICES 8 #define MAX_AGENT_ARGS_LEN 512 #define METHOD_NAME_PATH "/cluster/clusternodes/clusternode[@name=\"%s\"]/fence/method[%d]/@name"

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - fencing: Automatic detection of EOL during login process

by Marek Grác

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=b5253b0c741... Commit: b5253b0c74175ecda233f573fb0afa733b60eda2 Parent: 7d64a32254aac1c4bbeb7d7d684c606893d9fda1 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Mon Jul 23 17:23:39 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 17:23:39 2012 +0200 fencing: Automatic detection of EOL during login process Resolves: rhbz#810949 --- fence/agents/apc/fence_apc.py | 44 +++++++++++----------- fence/agents/bladecenter/fence_bladecenter.py | 18 +++++----- fence/agents/drac/fence_drac5.py | 12 +++--- fence/agents/ilo_mp/fence_ilo_mp.py | 12 +++--- fence/agents/lib/fencing.py.py | 48 +++++++++++++++--------- fence/agents/rsa/fence_rsa.py | 6 ++-- fence/agents/sanbox2/fence_sanbox2.py | 32 ++++++++-------- fence/agents/wti/fence_wti.py | 2 +- 8 files changed, 93 insertions(+), 81 deletions(-) diff --git a/fence/agents/apc/fence_apc.py b/fence/agents/apc/fence_apc.py index 0a3ce58..4125d39 100755 --- a/fence/agents/apc/fence_apc.py +++ b/fence/agents/apc/fence_apc.py @@ -29,7 +29,7 @@ def get_power_status(conn, options): exp_result = 0 outlets = {} try: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) version = 0 @@ -58,15 +58,15 @@ def get_power_status(conn, options): if switch == 0: if version == 2: if admin == 0: - conn.send("2\r\n") + conn.send_eol("2") else: - conn.send("3\r\n") + conn.send_eol("3") else: - conn.send("2\r\n") + conn.send_eol("2") conn.log_expect(options, options["-c"], int(options["-Y"])) - conn.send("1\r\n") + conn.send_eol("1") else: - conn.send(options["-s"]+"\r\n") + conn.send_eol(options["-s"]) while True: exp_result = conn.log_expect(options, [ options["-c"], "Press <ENTER>" ], int(options["-Y"])) @@ -76,7 +76,7 @@ def get_power_status(conn, options): res = show_re.search(x) if (res != None): outlets[res.group(2)] = (res.group(3), res.group(4)) - conn.send("\r\n") + conn.send_eol("") if exp_result == 0: break conn.send(chr(03)) @@ -103,7 +103,7 @@ def set_power_status(conn, options): }[options["-o"]] try: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) version = 0 @@ -138,41 +138,41 @@ def set_power_status(conn, options): if switch == 0: if version == 2: if admin2 == 0: - conn.send("2\r\n") + conn.send_eol("2") else: - conn.send("3\r\n") + conn.send_eol("3") else: - conn.send("2\r\n") + conn.send_eol("2") conn.log_expect(options, options["-c"], int(options["-Y"])) if (None == re.compile('.*2- Outlet Restriction.*', re.IGNORECASE | re.S).match(conn.before)): admin3 = 0 else: admin3 = 1 - conn.send("1\r\n") + conn.send_eol("1") else: - conn.send(options["-s"] + "\r\n") + conn.send_eol(options["-s"]) while 1 == conn.log_expect(options, [ options["-c"], "Press <ENTER>" ], int(options["-Y"])): - conn.send("\r\n") - conn.send(options["-n"]+"\r\n") + conn.send_eol("") + conn.send_eol(options["-n"]+"") conn.log_expect(options, options["-c"], int(options["-Y"])) if switch == 0: if admin2 == 1: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) if admin3 == 1: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) else: - conn.send("1\r\n") + conn.send_eol("1") conn.log_expect(options, options["-c"], int(options["-Y"])) - conn.send(action+"\r\n") + conn.send_eol(action) conn.log_expect(options, "Enter 'YES' to continue or <ENTER> to cancel :", int(options["-Y"])) - conn.send("YES\r\n") + conn.send_eol("YES") conn.log_expect(options, "Press <ENTER> to continue...", int(options["-Y"])) - conn.send("\r\n") + conn.send_eol("") conn.log_expect(options, options["-c"], int(options["-Y"])) conn.send(chr(03)) conn.log_expect(options, "- Logout", int(options["-Y"])) @@ -231,7 +231,7 @@ will block any necessary fencing actions." ## a problem because everything is checked before. ###### try: - conn.sendline("4") + conn.send_eol("4") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/bladecenter/fence_bladecenter.py b/fence/agents/bladecenter/fence_bladecenter.py index ac65f65..d97b9e4 100755 --- a/fence/agents/bladecenter/fence_bladecenter.py +++ b/fence/agents/bladecenter/fence_bladecenter.py @@ -27,7 +27,7 @@ def get_power_status(conn, options): try: node_cmd = "system:blade\[" + options["-n"] + "\]>" - conn.send("env -T system:blade[" + options["-n"] + "]\r\n") + conn.send_eol("env -T system:blade[" + options["-n"] + "]") i = conn.log_expect(options, [ node_cmd, "system>" ] , int(options["-Y"])) if i == 1: ## Given blade number does not exist @@ -35,10 +35,10 @@ def get_power_status(conn, options): return "off" else: fail(EC_STATUS) - conn.send("power -state\r\n") + conn.send_eol("power -state") conn.log_expect(options, node_cmd, int(options["-Y"])) status = conn.before.splitlines()[-1] - conn.send("env -T system\r\n") + conn.send_eol("env -T system") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -56,7 +56,7 @@ def set_power_status(conn, options): try: node_cmd = "system:blade\[" + options["-n"] + "\]>" - conn.send("env -T system:blade[" + options["-n"] + "]\r\n") + conn.send_eol("env -T system:blade[" + options["-n"] + "]") i = conn.log_expect(options, [ node_cmd, "system>" ] , int(options["-Y"])) if i == 1: ## Given blade number does not exist @@ -65,9 +65,9 @@ def set_power_status(conn, options): else: fail(EC_GENERIC_ERROR) - conn.send("power -"+options["-o"]+"\r\n") + conn.send_eol("power -"+options["-o"]) conn.log_expect(options, node_cmd, int(options["-Y"])) - conn.send("env -T system\r\n") + conn.send_eol("env -T system") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -79,9 +79,9 @@ def get_blades_list(conn, options): try: node_cmd = "system>" - conn.send("env -T system\r\n") + conn.send_eol("env -T system") conn.log_expect(options, node_cmd, int(options["-Y"])) - conn.send("list -l 2\r\n") + conn.send_eol("list -l 2") conn.log_expect(options, node_cmd, int(options["-Y"])) lines = conn.before.split("\r\n") @@ -131,7 +131,7 @@ and uses the command line interface to power on and off blades." ## Logout from system ###### try: - conn.send("exit\r\n") + conn.send_eol("exit") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/drac/fence_drac5.py b/fence/agents/drac/fence_drac5.py index 23cd7ca..298339f 100755 --- a/fence/agents/drac/fence_drac5.py +++ b/fence/agents/drac/fence_drac5.py @@ -25,9 +25,9 @@ BUILD_DATE="" def get_power_status(conn, options): try: if options["model"] == "DRAC CMC": - conn.sendline("racadm serveraction powerstatus -m " + options["-m"]) + conn.send_eol("racadm serveraction powerstatus -m " + options["-m"]) elif options["model"] == "DRAC 5": - conn.sendline("racadm serveraction powerstatus") + conn.send_eol("racadm serveraction powerstatus") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: @@ -49,9 +49,9 @@ def set_power_status(conn, options): try: if options["model"] == "DRAC CMC": - conn.sendline("racadm serveraction " + action + " -m " + options["-m"]) + conn.send_eol("racadm serveraction " + action + " -m " + options["-m"]) elif options["model"] == "DRAC 5": - conn.sendline("racadm serveraction " + action) + conn.send_eol("racadm serveraction " + action) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -63,7 +63,7 @@ def get_list_devices(conn, options): try: if options["model"] == "DRAC CMC": - conn.sendline("getmodinfo") + conn.send_eol("getmodinfo") list_re = re.compile("^([^\s]*?)\s+Present\s*(ON|OFF)\s*.*$") conn.log_expect(options, options["-c"], int(options["-g"])) @@ -133,7 +133,7 @@ By default, the telnet interface is not enabled." ## Logout from system ###### try: - conn.sendline("exit") + conn.send_eol("exit") time.sleep(1) conn.close() except exceptions.OSError: diff --git a/fence/agents/ilo_mp/fence_ilo_mp.py b/fence/agents/ilo_mp/fence_ilo_mp.py index 06e1358..40ac597 100644 --- a/fence/agents/ilo_mp/fence_ilo_mp.py +++ b/fence/agents/ilo_mp/fence_ilo_mp.py @@ -12,7 +12,7 @@ BUILD_DATE="" def get_power_status(conn, options): try: - conn.send("show /system1\r\n") + conn.send_eol("show /system1") re_state = re.compile('EnabledState=(.*)', re.IGNORECASE) conn.log_expect(options, re_state, int(options["-Y"])) @@ -31,9 +31,9 @@ def get_power_status(conn, options): def set_power_status(conn, options): try: if options["-o"] == "on": - conn.send("start /system1\r\n") + conn.send_eol("start /system1") else: - conn.send("stop -f /system1\r\n") + conn.send_eol("stop -f /system1") conn.log_expect(options, options["-c"], int(options["-g"])) @@ -46,7 +46,7 @@ def set_power_status(conn, options): def main(): device_opt = [ "help", "version", "agent", "quiet", "verbose", "debug", "action", "ipaddr", "login", "passwd", "passwd_script", - "secure", "identity_file", "cmd_prompt", "ipport", "login_eol_lf", + "secure", "identity_file", "cmd_prompt", "ipport", "separator", "inet4_only", "inet6_only", "power_timeout", "shell_timeout", "login_timeout", "power_wait" ] @@ -64,7 +64,7 @@ def main(): show_docs(options, docs) conn = fence_login(options) - conn.send("SMCLP\r\n") + conn.send_eol("SMCLP") ## ## Fence operations @@ -72,7 +72,7 @@ def main(): result = fence_action(conn, options, set_power_status, get_power_status) try: - conn.send("exit\r\n") + conn.send_eol("exit") except exceptions.OSError: pass except pexpect.ExceptionPexpect: diff --git a/fence/agents/lib/fencing.py.py b/fence/agents/lib/fencing.py.py index c4fe218..25a0f7e 100644 --- a/fence/agents/lib/fencing.py.py +++ b/fence/agents/lib/fencing.py.py @@ -178,11 +178,6 @@ all_opt = { "required" : "0", "shortdesc" : "Force ribcl version to use", "order" : 1 }, - "login_eol_lf" : { - "getopt" : "", - "help" : "", - "order" : 1 - }, "cmd_prompt" : { "getopt" : "c:", "longopt" : "command-prompt", @@ -403,12 +398,20 @@ all_opt = { common_opt = [ "retry_on", "delay" ] class fspawn(pexpect.spawn): + def __init__(self, options, command): + pexpect.spawn.__init__(self, command) + self.opt = options + def log_expect(self, options, pattern, timeout): result = self.expect(pattern, timeout) if options["log"] >= LOG_MODE_VERBOSE: options["debug_fh"].write(self.before + self.after) return result + # send EOL according to what was detected in login process (telnet) + def send_eol(self, message): + self.send(message + self.opt["eol"]) + def atexit_handler(): try: sys.stdout.close() @@ -859,10 +862,7 @@ def fence_login(options): if (options.has_key("-4")): force_ipvx="-4 " - if (options["device_opt"].count("login_eol_lf")): - login_eol = "\n" - else: - login_eol = "\r\n" + options["eol"] = "\r\n" ## Do the delay of the fence device before logging in ## Delay is important for two-node clusters fencing but we do not need to delay 'status' operations @@ -876,7 +876,7 @@ def fence_login(options): if options.has_key("-z"): command = '%s %s %s %s' % (SSL_PATH, force_ipvx, options["-a"], options["-u"]) try: - conn = fspawn(command) + conn = fspawn(options, command) except pexpect.ExceptionPexpect, ex: ## SSL telnet is part of the fencing package sys.stderr.write(str(ex) + "\n") @@ -886,7 +886,7 @@ def fence_login(options): if options.has_key("ssh_options"): command += ' ' + options["ssh_options"] try: - conn = fspawn(command) + conn = fspawn(options, command) except pexpect.ExceptionPexpect, ex: sys.stderr.write(str(ex) + "\n") sys.stderr.write("Due to limitations, binary dependencies on fence agents " @@ -915,7 +915,7 @@ def fence_login(options): if options.has_key("ssh_options"): command += ' ' + options["ssh_options"] try: - conn = fspawn(command) + conn = fspawn(options, command) except pexpect.ExceptionPexpect, ex: sys.stderr.write(str(ex) + "\n") sys.stderr.write("Due to limitations, binary dependencies on fence agents " @@ -934,7 +934,7 @@ def fence_login(options): fail_usage("Failed: You have to enter passphrase (-p) for identity file") else: try: - conn = fspawn(TELNET_PATH) + conn = fspawn(options, TELNET_PATH) conn.send("set binary\n") conn.send("open %s -%s\n"%(options["-a"], options["-u"])) except pexpect.ExceptionPexpect, ex: @@ -943,11 +943,23 @@ def fence_login(options): "are not in the spec file and must be installed separately." + "\n") sys.exit(EC_GENERIC_ERROR) - conn.log_expect(options, re_login, int(options["-y"])) - conn.send(options["-l"] + login_eol) - conn.log_expect(options, re_pass, int(options["-Y"])) - conn.send(options["-p"] + login_eol) - conn.log_expect(options, options["-c"], int(options["-Y"])) + result = conn.log_expect(options, re_login, int(options["-y"])) + conn.send_eol(options["-l"]) + + ## automatically change end of line separator + screen = conn.read_nonblocking(size=100, timeout=int(options["-Y"])) + if (re_login.search(screen) != None): + options["eol"] = "\n" + conn.send_eol(options["-l"]) + result = conn.log_expect(options, re_pass, int(options["-y"])) + elif (re_pass.search(screen) == None): + conn.log_expect(options, re_pass, int(options["-Y"])) + + try: + conn.send_eol(options["-p"]) + conn.log_expect(options, options["-c"], int(options["-Y"])) + except KeyError: + fail(EC_PASSWORD_MISSING) except pexpect.EOF: fail(EC_LOGIN_DENIED) except pexpect.TIMEOUT: diff --git a/fence/agents/rsa/fence_rsa.py b/fence/agents/rsa/fence_rsa.py index 33c7fe9..d82c448 100755 --- a/fence/agents/rsa/fence_rsa.py +++ b/fence/agents/rsa/fence_rsa.py @@ -19,7 +19,7 @@ BUILD_DATE="March, 2009" def get_power_status(conn, options): try: - conn.send("power state\r\n") + conn.send_eol("power state") conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -36,7 +36,7 @@ def get_power_status(conn, options): def set_power_status(conn, options): try: - conn.send("power " + options["-o"] + "\r\n") + conn.send_eol("power " + options["-o"]) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) @@ -84,7 +84,7 @@ will block any necessary fencing actions." ## Logout from system ###### try: - conn.sendline("exit") + conn.send_eol("exit") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/sanbox2/fence_sanbox2.py b/fence/agents/sanbox2/fence_sanbox2.py index c95e602..d20f7e2 100644 --- a/fence/agents/sanbox2/fence_sanbox2.py +++ b/fence/agents/sanbox2/fence_sanbox2.py @@ -24,14 +24,14 @@ def get_power_status(conn, options): 'offline' : "off" } try: - conn.send("show port " + options["-n"] + "\n") + conn.send_eol("show port " + options["-n"]) conn.log_expect(options, options["-c"], int(options["-Y"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass @@ -51,28 +51,28 @@ def set_power_status(conn, options): }[options["-o"]] try: - conn.send("set port " + options["-n"] + " state " + action + "\n") + conn.send_eol("set port " + options["-n"] + " state " + action) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass fail(EC_TIMED_OUT) try: - conn.send("set port " + options["-n"] + " state " + action + "\n") + conn.send_eol("set port " + options["-n"] + " state " + action) conn.log_expect(options, options["-c"], int(options["-g"])) except pexpect.EOF: fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass @@ -82,7 +82,7 @@ def get_list_devices(conn, options): outlets = { } try: - conn.send("show port" + "\n") + conn.send_eol("show port") conn.log_expect(options, options["-c"], int(options["-Y"])) list_re = re.compile("^\s+(\d+?)\s+(Online|Offline)\s+", re.IGNORECASE) @@ -98,8 +98,8 @@ def get_list_devices(conn, options): fail(EC_CONNECTION_LOST) except pexpect.TIMEOUT: try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit") conn.close() except: pass @@ -110,7 +110,7 @@ def get_list_devices(conn, options): def main(): device_opt = [ "help", "version", "agent", "quiet", "verbose", "debug", "io_fencing", "ipaddr", "login", "passwd", "passwd_script", - "cmd_prompt", "port", "ipport", "login_eol_lf", "separator", + "cmd_prompt", "port", "ipport", "separator", "power_timeout", "shell_timeout", "login_timeout", "power_wait" ] atexit.register(atexit_handler) @@ -138,7 +138,7 @@ because the connection will block any necessary fencing actions." ## conn = fence_login(options) - conn.send("admin start\n") + conn.send_eol("admin start") conn.log_expect(options, options["-c"], int(options["-Y"])) if (re.search("$admin$", conn.before, re.MULTILINE) == None): @@ -153,8 +153,8 @@ because the connection will block any necessary fencing actions." ## Logout from system ###### try: - conn.send("admin end\n") - conn.send("exit\n") + conn.send_eol("admin end") + conn.send_eol("exit\n") conn.close() except exceptions.OSError: pass diff --git a/fence/agents/wti/fence_wti.py b/fence/agents/wti/fence_wti.py index c742e71..40b97e3 100755 --- a/fence/agents/wti/fence_wti.py +++ b/fence/agents/wti/fence_wti.py @@ -125,7 +125,7 @@ is running because the connection will block any necessary fencing actions." if 0 == options.has_key("-x"): try: try: - conn = fspawn(TELNET_PATH) + conn = fspawn(options, TELNET_PATH) conn.send("set binary\n") conn.send("open %s -%s\n"%(options["-a"], options["-u"])) except pexpect.ExceptionPexpect, ex:

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - fence_apc: fence_apc fails for some port numbers

by Marek Grác

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=7d64a32254a... Commit: 7d64a32254aac1c4bbeb7d7d684c606893d9fda1 Parent: 1ed8b72936fa7d871dbdc9b787e7e775d1ea3699 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Wed Jun 23 18:29:02 2010 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 17:20:28 2012 +0200 fence_apc: fence_apc fails for some port numbers Problem is that string we parse looks like: 'Press Enter to continue ...\r 23-bar1 ON' and we did not expect text message before our line. Resolves: #606315 --- fence/agents/apc/fence_apc.py | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/fence/agents/apc/fence_apc.py b/fence/agents/apc/fence_apc.py index 1aa7bed..0a3ce58 100755 --- a/fence/agents/apc/fence_apc.py +++ b/fence/agents/apc/fence_apc.py @@ -71,11 +71,11 @@ def get_power_status(conn, options): while True: exp_result = conn.log_expect(options, [ options["-c"], "Press <ENTER>" ], int(options["-Y"])) lines = conn.before.split("\n"); - show_re = re.compile('^\s*(\d+)- (.*?)\s+(ON|OFF)\s*') + show_re = re.compile('(^|\x0D)\s*(\d+)- (.*?)\s+(ON|OFF)\s*') for x in lines: res = show_re.search(x) if (res != None): - outlets[res.group(1)] = (res.group(2), res.group(3)) + outlets[res.group(2)] = (res.group(3), res.group(4)) conn.send("\r\n") if exp_result == 0: break

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - fencing: Add iPDU fencing agent for model 46M4002

by Marek Grác

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=1ed8b72936f... Commit: 1ed8b72936fa7d871dbdc9b787e7e775d1ea3699 Parent: 76741bb2a94ae94e493c609d50f570d02e2f3029 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Mon Jul 23 17:06:32 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 17:06:32 2012 +0200 fencing: Add iPDU fencing agent for model 46M4002 Resolves: rhbz#741985 --- fence/agents/Makefile | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/fence/agents/Makefile b/fence/agents/Makefile index 716530c..669e291 100644 --- a/fence/agents/Makefile +++ b/fence/agents/Makefile @@ -30,6 +30,7 @@ all: ${MAKE} -C ifmib all ${MAKE} -C ilo all ${MAKE} -C ilo_mp all + ${MAKE} -C ipdu all ${MAKE} -C ipmilan all ${MAKE} -C lpar all ${MAKE} -C manual all @@ -65,6 +66,7 @@ install: all ${MAKE} -C ifmib install ${MAKE} -C ilo install ${MAKE} -C ilo_mp install + ${MAKE} -C ipdu install ${MAKE} -C ipmilan install ${MAKE} -C lpar install ${MAKE} -C manual install @@ -100,6 +102,7 @@ clean: ${MAKE} -C ifmib clean ${MAKE} -C ilo clean ${MAKE} -C ilo_mp clean + ${MAKE} -C ipdu clean ${MAKE} -C ipmilan clean ${MAKE} -C lpar clean ${MAKE} -C manual clean

11 years, 9 months

1
0
0 / 0

cluster: RHEL59 - qdiskd: Make multipath issues go away

by Fabio M. Di Nitto

Gitweb: http://git.fedorahosted.org/git/?p=cluster.git;a=commitdiff;h=76741bb2a94... Commit: 76741bb2a94ae94e493c609d50f570d02e2f3029 Parent: 3290dff9b88992e913b28dd57c6b64a3c2b75c8a Author: Lon Hohberger <lhh(a)redhat.com> AuthorDate: Fri May 6 10:14:04 2011 -0400 Committer: Fabio M. Di Nitto <fdinitto(a)redhat.com> CommitterDate: Mon Jul 23 14:54:07 2012 +0200 qdiskd: Make multipath issues go away Qdiskd hsitorically has required significant tuning to work around delays which occur during multipath failover, overloaded I/O, and LUN trespasses in both device-mapper-multipath and EMC PowerPath environments. This patch goes a very long way towards eliminating false evictions when these conditions occur by making qdiskd whine to the other cluster members when it detects hung system calls. When a cluster member whines, it indicates the source of the problem (which system call is hung), and the act of receiving a whine from a host indicates that qdiskd is operational, but that I/O is hung. Hung I/O is different from losing storage entirely (where you get I/O errors). Possible problems: - Receive queue getting very full, causing messages to become blocked on a node where I/O is hung. 1) that would take a very long time, and 2) node should get evicted at that point anyway. Resolves: rhbz#782900 this version of the patch is a backport of: e2937eb33f224f86904fead08499a6178868ca6a 34d2872fb7e60be1594158acaaeb8acd74f78d22 There is a minor change vs original patch based on how qdiskd in RHEL5 handles cman connection. We add an extra call to cman_alive in main qdisk_loop to make sure data are not stalled on the cman port, and data_callback to qdiskd_whine executed. Signed-off-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Fabio M. Di Nitto <fdinitto(a)redhat.com> --- cman/daemon/cnxman-socket.h | 1 + cman/qdisk/Makefile | 2 +- cman/qdisk/disk.h | 6 ++++ cman/qdisk/iostate.c | 17 +++++++++++-- cman/qdisk/iostate.h | 4 ++- cman/qdisk/main.c | 54 +++++++++++++++++++++++++++++++++++++++---- 6 files changed, 74 insertions(+), 10 deletions(-) diff --git a/cman/daemon/cnxman-socket.h b/cman/daemon/cnxman-socket.h index 351c97c..1d01b44 100644 --- a/cman/daemon/cnxman-socket.h +++ b/cman/daemon/cnxman-socket.h @@ -79,6 +79,7 @@ #define CLUSTER_PORT_SERVICES 2 #define CLUSTER_PORT_SYSMAN 10 /* Remote execution daemon */ #define CLUSTER_PORT_CLVMD 11 /* Cluster LVM daemon */ +#define CLUSTER_PORT_QDISKD 178 /* Quorum disk daemon */ /* Port numbers above this will be blocked when the cluster is inquorate or in * transition */ diff --git a/cman/qdisk/Makefile b/cman/qdisk/Makefile index f58806b..9bfc486 100644 --- a/cman/qdisk/Makefile +++ b/cman/qdisk/Makefile @@ -32,7 +32,7 @@ qdiskd: disk.o crc32.o disk_util.o main.o score.o bitmap.o clulog.o \ gcc -o $@ $^ -lpthread -L../lib -L${ccslibdir} -lccs -lrt mkqdisk: disk.o crc32.o disk_util.o iostate.o \ - proc.o mkqdisk.o scandisk.o clulog.o gettid.o + proc.o mkqdisk.o scandisk.o clulog.o gettid.o ../lib/libcman.a gcc -o $@ $^ -lrt %.o: %.c diff --git a/cman/qdisk/disk.h b/cman/qdisk/disk.h index b784220..d491de1 100644 --- a/cman/qdisk/disk.h +++ b/cman/qdisk/disk.h @@ -290,6 +290,12 @@ typedef struct { status_block_t ni_status; } node_info_t; +typedef struct { + qd_ctx *ctx; + node_info_t *ni; + size_t ni_len; +} qd_priv_t; + int qd_write_status(qd_ctx *ctx, int nid, disk_node_state_t state, disk_msg_t *msg, memb_mask_t mask, memb_mask_t master); int qd_read_print_status(target_info_t *disk, int nid); diff --git a/cman/qdisk/iostate.c b/cman/qdisk/iostate.c index 65b4d50..eb74ad2 100644 --- a/cman/qdisk/iostate.c +++ b/cman/qdisk/iostate.c @@ -1,10 +1,14 @@ #include <pthread.h> +#include <libcman.h> #include <iostate.h> #include <unistd.h> #include <time.h> #include <sys/time.h> #include <clulog.h> +#include <stdint.h> +#include "platform.h" #include "iostate.h" +#include "../daemon/cnxman-socket.h" static iostate_t main_state = 0; static int main_incarnation = 0; @@ -26,7 +30,7 @@ static struct state_table io_state_table[] = { { STATE_LSEEK, "seek" }, { -1, NULL } }; -static const char * +const char * state_to_string(iostate_t state) { static const char *ret = "unknown"; @@ -65,6 +69,8 @@ io_nanny_thread(void *arg) iostate_t last_main_state = 0, current_main_state = 0; int last_main_incarnation = 0, current_main_incarnation = 0; int logged_incarnation = 0; + cman_handle_t ch = (cman_handle_t)arg; + int32_t whine_state; /* Start with wherever we're at now */ pthread_mutex_lock(&state_mutex); @@ -96,6 +102,11 @@ io_nanny_thread(void *arg) continue; } + /* Whine on CMAN api */ + whine_state = (int32_t)current_main_state; + swab32(whine_state); + cman_send_data(ch, &whine_state, sizeof(int32_t), 0, CLUSTER_PORT_QDISKD, 0); + /* Don't log things twice */ if (logged_incarnation == current_main_incarnation) continue; @@ -114,7 +125,7 @@ io_nanny_thread(void *arg) int -io_nanny_start(int timeout) +io_nanny_start(cman_handle_t ch, int timeout) { int ret; @@ -124,7 +135,7 @@ io_nanny_start(int timeout) qdisk_timeout = timeout; thread_active = 1; - ret = pthread_create(&io_nanny_tid, NULL, io_nanny_thread, NULL); + ret = pthread_create(&io_nanny_tid, NULL, io_nanny_thread, ch); pthread_mutex_unlock(&state_mutex); return ret; diff --git a/cman/qdisk/iostate.h b/cman/qdisk/iostate.h index 7dd7bf6..a65b1d4 100644 --- a/cman/qdisk/iostate.h +++ b/cman/qdisk/iostate.h @@ -11,7 +11,9 @@ typedef enum { void io_state(iostate_t state); -int io_nanny_start(int timeout); +int io_nanny_start(cman_handle_t ch, int timeout); int io_nanny_stop(void); +const char * state_to_string(iostate_t state); + #endif diff --git a/cman/qdisk/main.c b/cman/qdisk/main.c index 0d7bb3d..90d00ab 100644 --- a/cman/qdisk/main.c +++ b/cman/qdisk/main.c @@ -48,6 +48,7 @@ (defined(LIBCMAN_VERSION) && LIBCMAN_VERSION < 2)) #include <cluster/cnxman-socket.h> #endif +#include "../daemon/cnxman-socket.h" int daemon_init(char *); int check_process_running(char *, pid_t *); @@ -892,6 +893,11 @@ quorum_loop(qd_ctx *ctx, node_info_t *ni, int max) _running = 1; while (_running) { + /* perform a forceful cman dispatch */ + if (cman_alive(ctx->qc_ch) < 0) { + clulog(LOG_ERR, "cman: %s\n", strerror(errno)); + } + /* XXX this was getuptime() in clumanager */ get_time(&oldtime, (ctx->qc_flags&RF_UPTIME)); @@ -1514,6 +1520,31 @@ check_stop_cman(qd_ctx *ctx) } } +static void +qdisk_whine(cman_handle_t h, void *privdata, char *buf, int len, + uint8_t port, int nodeid) +{ + int32_t dstate; + qd_priv_t *qp = (qd_priv_t *)privdata; + node_info_t *ni = qp->ni; + + if (len != sizeof(dstate)) { + return; + } + + dstate = *((int32_t*)buf); + + if (nodeid == (qp->ctx->qc_my_id)) + return; + + swab32(dstate); + + if (dstate) { + clulog(LOG_CRIT, "qdiskd on node %d reports hung %s()\n", + state_to_string(dstate)); + ni[nodeid-1].ni_misses = 0; + } +} int main(int argc, char **argv) @@ -1528,6 +1559,7 @@ main(int argc, char **argv) char device[128]; pid_t pid; quorum_header_t qh; + qd_priv_t qp; if (check_process_running(argv[0], &pid) && pid !=getpid()) { printf("QDisk services already running\n"); @@ -1559,10 +1591,16 @@ main(int argc, char **argv) } } + /* For cman notifications we need two sockets - one for events, + one for config change callbacks */ + qp.ctx = &ctx; + qp.ni = &ni[0]; + qp.ni_len = MAX_NODES_DISK; + #if (defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2) - ch = cman_admin_init(NULL); + ch = cman_admin_init(&qp); #else - ch = cman_init(NULL); + ch = cman_init(&qp); #endif if (!ch) { if (!foreground && !forked) { @@ -1577,13 +1615,19 @@ main(int argc, char **argv) do { sleep(5); #if (defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2) - ch = cman_admin_init(NULL); + ch = cman_admin_init(&qp); #else - ch = cman_init(NULL); + ch = cman_init(&qp); #endif } while (!ch); } + if (cman_start_recv_data(ch, qdisk_whine, CLUSTER_PORT_QDISKD) != 0) { + clulog_and_print(LOG_CRIT, "Could not register with CMAN: %s\n", + strerror(errno)); + goto out; + } + memset(&me, 0, sizeof(me)); while (cman_get_node(ch, CMAN_NODEID_US, &me) < 0) { if (!foreground && !forked) { @@ -1696,7 +1740,7 @@ main(int argc, char **argv) } } - io_nanny_start(ctx.qc_tko * ctx.qc_interval); + io_nanny_start(ch, ctx.qc_tko * ctx.qc_interval); if (quorum_loop(&ctx, ni, MAX_NODES_DISK) == 0) cman_unregister_quorum_device(ctx.qc_ch);

11 years, 9 months

1
0
0 / 0

fence-agents: master - fence_ipdu: Minor fixes to fence agent

by Marek Grác

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=dff6b1... Commit: dff6b12ab73c8c38fc579ec312f398d6711cfeb8 Parent: 74d1242d8b968d5886be4394c8792d8c8d93f321 Author: Marek 'marx' Grac <mgrac(a)redhat.com> AuthorDate: Mon Jul 23 11:24:43 2012 +0200 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 11:24:43 2012 +0200 fence_ipdu: Minor fixes to fence agent * replace code to use 'default' value * remove code for backward compatibility with old options (-n switch:port) Resolves: rhbz#741985 --- fence/agents/ipdu/fence_ipdu.py | 13 ++----------- 1 files changed, 2 insertions(+), 11 deletions(-) diff --git a/fence/agents/ipdu/fence_ipdu.py b/fence/agents/ipdu/fence_ipdu.py index 9c135e6..aac0b7d 100644 --- a/fence/agents/ipdu/fence_ipdu.py +++ b/fence/agents/ipdu/fence_ipdu.py @@ -49,7 +49,7 @@ def ipdu_set_device(conn,options): agents_dir={'.1.3.6.1.4.1.2.6.223':IBMiPDU, None:IBMiPDU} - # First resolve type of APC + # First resolve type of PDU device pdu_type=conn.walk(OID_SYS_OBJECT_ID) if (not ((len(pdu_type)==1) and (agents_dir.has_key(pdu_type[0][1])))): @@ -133,6 +133,7 @@ def get_outlets_status(conn, options): def ipdu_snmp_define_defaults(): all_opt["snmp_version"]["default"]="3" all_opt["community"]["default"]="private" + all_opt["switch"]["default"]="1" device=IBMiPDU # Main agent method @@ -152,16 +153,6 @@ def main(): options=check_input(device_opt,process_input(device_opt)) - ## Support for -n [switch]:[plug] notation that was used before - if ((options.has_key("-n")) and (-1 != options["-n"].find(":"))): - (switch, plug) = options["-n"].split(":", 1) - if ((switch.isdigit()) and (plug.isdigit())): - options["-s"] = switch - options["-n"] = plug - - if (not (options.has_key("-s"))): - options["-s"]="1" - docs = { } docs["shortdesc"] = "Fence agent for iPDU over SNMP" docs["longdesc"] = "fence_ipdu is an I/O Fencing agent \

11 years, 9 months

1
0
0 / 0

fence-agents: master - Add iPDU fencing agent for model 46M4002

by Marek Grác

Gitweb: http://git.fedorahosted.org/git/?p=fence-agents.git;a=commitdiff;h=74d124... Commit: 74d1242d8b968d5886be4394c8792d8c8d93f321 Parent: efcf34f5156385c397509f9fe40ae08ca74a96a0 Author: Lon Hohberger <lhh(a)redhat.com> AuthorDate: Fri Jul 20 15:43:14 2012 -0400 Committer: Marek 'marx' Grac <mgrac(a)redhat.com> CommitterDate: Mon Jul 23 11:10:19 2012 +0200 Add iPDU fencing agent for model 46M4002 Resolves: rhbz#741985 Signed-off-by: Lon Hohberger <lhh(a)redhat.com> Signed-off-by: Marek 'marx' Grac <mgrac(a)redhat.com> --- .gitignore | 1 + configure.ac | 1 + fence/agents/ipdu/Makefile.am | 17 ++++ fence/agents/ipdu/fence_ipdu.py | 179 +++++++++++++++++++++++++++++++++++++++ 4 files changed, 198 insertions(+), 0 deletions(-) diff --git a/.gitignore b/.gitignore index 3a182d8..c1b028c 100644 --- a/.gitignore +++ b/.gitignore @@ -49,6 +49,7 @@ fence/agents/ifmib/fence_ifmib fence/agents/ilo/fence_ilo fence/agents/ilo_mp/fence_ilo_mp fence/agents/intelmodular/fence_intelmodular +fence/agents/ipdu/fence_ipdu fence/agents/ipmilan/fence_ipmilan fence/agents/ldom/fence_ldom fence/agents/lib/fencing.py diff --git a/configure.ac b/configure.ac index 4797740..1b0ee22 100644 --- a/configure.ac +++ b/configure.ac @@ -263,6 +263,7 @@ AC_CONFIG_FILES([Makefile fence/agents/eps/Makefile fence/agents/hpblade/Makefile fence/agents/ibmblade/Makefile + fence/agents/ipdu/Makefile fence/agents/ifmib/Makefile fence/agents/ilo/Makefile fence/agents/ilo_mp/Makefile diff --git a/fence/agents/ipdu/Makefile.am b/fence/agents/ipdu/Makefile.am new file mode 100644 index 0000000..fb02cc8 --- /dev/null +++ b/fence/agents/ipdu/Makefile.am @@ -0,0 +1,17 @@ +MAINTAINERCLEANFILES = Makefile.in + +TARGET = fence_ipdu + +SRC = $(TARGET).py + +EXTRA_DIST = $(SRC) + +sbin_SCRIPTS = $(TARGET) + +man_MANS = $(TARGET).8 + +include $(top_srcdir)/make/fencebuild.mk +include $(top_srcdir)/make/fenceman.mk + +clean-local: clean-man + rm -f $(TARGET) diff --git a/fence/agents/ipdu/fence_ipdu.py b/fence/agents/ipdu/fence_ipdu.py new file mode 100644 index 0000000..9c135e6 --- /dev/null +++ b/fence/agents/ipdu/fence_ipdu.py @@ -0,0 +1,179 @@ +#!/usr/bin/python + +# The Following agent has been tested on: +# IBM iPDU model 46M4002 +# Firmware release OPDP_sIBM_v01.2_1 +# + +import sys, re, pexpect +sys.path.append("/usr/share/fence") +from fencing import * +from fencing_snmp import * + +#BEGIN_VERSION_GENERATION +RELEASE_VERSION="IBM iPDU SNMP fence agent" +REDHAT_COPYRIGHT="" +BUILD_DATE="" +#END_VERSION_GENERATION + +### CONSTANTS ### +# oid defining fence device +OID_SYS_OBJECT_ID='.1.3.6.1.2.1.1.2.0' + +### GLOBAL VARIABLES ### +# Device - see IBM iPDU +device=None + +# Port ID +port_id=None +# Switch ID +switch_id=None + +# Classes describing Device params +class IBMiPDU: + # iPDU + status_oid= '.1.3.6.1.4.1.2.6.223.8.2.2.1.11.%d' + control_oid= '.1.3.6.1.4.1.2.6.223.8.2.2.1.11.%d' + outlet_table_oid='.1.3.6.1.4.1.2.6.223.8.2.2.1.2' + ident_str="IBM iPDU" + state_on=1 + state_off=0 + turn_on=1 + turn_off=0 + has_switches=False + +### FUNCTIONS ### +def ipdu_set_device(conn,options): + global device + + agents_dir={'.1.3.6.1.4.1.2.6.223':IBMiPDU, + None:IBMiPDU} + + # First resolve type of APC + pdu_type=conn.walk(OID_SYS_OBJECT_ID) + + if (not ((len(pdu_type)==1) and (agents_dir.has_key(pdu_type[0][1])))): + pdu_type=[[None,None]] + + device=agents_dir[pdu_type[0][1]] + + conn.log_command("Trying %s"%(device.ident_str)) + +def ipdu_resolv_port_id(conn,options): + global port_id,switch_id,device + + if (device==None): + ipdu_set_device(conn,options) + + # Now we resolv port_id/switch_id + if ((options["-n"].isdigit()) and ((not device.has_switches) or (options["-s"].isdigit()))): + port_id=int(options["-n"]) + + if (device.has_switches): + switch_id=int(options["-s"]) + else: + table=conn.walk(device.outlet_table_oid,30) + + for x in table: + if (x[1].strip('"')==options["-n"]): + t=x[0].split('.') + if (device.has_switches): + port_id=int(t[len(t)-1]) + switch_id=int(t[len(t)-3]) + else: + port_id=int(t[len(t)-1]) + + if (port_id==None): + fail_usage("Can't find port with name %s!"%(options["-n"])) + +def get_power_status(conn,options): + global port_id,switch_id,device + + if (port_id==None): + ipdu_resolv_port_id(conn,options) + + oid=((device.has_switches) and device.status_oid%(switch_id,port_id) or device.status_oid%(port_id)) + + (oid,status)=conn.get(oid) + return (status==str(device.state_on) and "on" or "off") + +def set_power_status(conn, options): + global port_id,switch_id,device + + if (port_id==None): + ipdu_resolv_port_id(conn,options) + + oid=((device.has_switches) and device.control_oid%(switch_id,port_id) or device.control_oid%(port_id)) + + conn.set(oid,(options["-o"]=="on" and device.turn_on or device.turn_off)) + + +def get_outlets_status(conn, options): + global device + + result={} + + if (device==None): + ipdu_set_device(conn,options) + + res_ports=conn.walk(device.outlet_table_oid,30) + + for x in res_ports: + t=x[0].split('.') + + port_num=((device.has_switches) and "%s:%s"%(t[len(t)-3],t[len(t)-1]) or "%s"%(t[len(t)-1])) + + port_name=x[1].strip('"') + port_status="" + result[port_num]=(port_name,port_status) + + return result + +# Define new options +def ipdu_snmp_define_defaults(): + all_opt["snmp_version"]["default"]="3" + all_opt["community"]["default"]="private" + device=IBMiPDU + +# Main agent method +def main(): + device_opt = [ "help", "version", "agent", "quiet", "verbose", "debug", + "action", "ipaddr", "login", "passwd", "passwd_script", + "test", "port", "separator", "no_login", "no_password", + "snmp_version", "community", "snmp_auth_prot", "snmp_sec_level", + "snmp_priv_prot", "snmp_priv_passwd", "snmp_priv_passwd_script", + "udpport","inet4_only","inet6_only", + "power_timeout", "shell_timeout", "login_timeout", "power_wait" ] + + atexit.register(atexit_handler) + + snmp_define_defaults () + ipdu_snmp_define_defaults() + + options=check_input(device_opt,process_input(device_opt)) + + ## Support for -n [switch]:[plug] notation that was used before + if ((options.has_key("-n")) and (-1 != options["-n"].find(":"))): + (switch, plug) = options["-n"].split(":", 1) + if ((switch.isdigit()) and (plug.isdigit())): + options["-s"] = switch + options["-n"] = plug + + if (not (options.has_key("-s"))): + options["-s"]="1" + + docs = { } + docs["shortdesc"] = "Fence agent for iPDU over SNMP" + docs["longdesc"] = "fence_ipdu is an I/O Fencing agent \ +which can be used with the IBM iPDU network power switch. It logs \ +into a device via SNMP and reboots a specified outlet. It supports \ +SNMP v3 with all combinations of authenticity/privacy settings." + docs["vendorurl"] = "http://www.ibm.com" + show_docs(options, docs) + + # Operate the fencing device + result = fence_action(FencingSnmp(options), options, set_power_status, get_power_status, get_outlets_status) + + sys.exit(result) +if __name__ == "__main__": + main()

11 years, 9 months

1
0
0 / 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

cluster-commits July 2012