From 7b21619f0022b9eba7ad5b9c3c018471c4be9a93 Mon Sep 17 00:00:00 2001

From: Paul Evans <pevans@redhat.com>

Date: Mon, 8 Feb 2021 18:14:37 +0000

Subject: [PATCH 1/7] pmdahacluster: Quick README.md update

Update README.md to reflect pmns root node update to ha_cluster

---

 src/pmdas/hacluster/README.md | 4 ++--

 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/pmdas/hacluster/README.md b/src/pmdas/hacluster/README.md

index 4d0fcabf5..709cb721d 100644

--- a/src/pmdas/hacluster/README.md

+++ b/src/pmdas/hacluster/README.md

@@ -6,7 +6,7 @@ The PMDA collects it's metric data from the following components that make up a

 ## General Notes

-### `hacluster.drbd.split_brain`

+### `ha_cluster.drbd.split_brain`

 This metric signals if there is a split brain occurring in DRBD per instance resource:volume. The metric will return the value `1` if a split brain is detected, otherwise it will be `0`.

@@ -37,7 +37,7 @@ exposed by this PMDA.

 Once the PMDA has been installed, the following command will list all of

 the available metrics and their explanatory “help” text:

-		# $ pminfo -fT hacluster

+		# $ pminfo -fT ha_cluster

 ## Installation

-- 

2.31.1

From f68df957286df0b0c2bb091d1025cf3c4adc2810 Mon Sep 17 00:00:00 2001

From: Ken McDonell <kenj@kenj.id.au>

Date: Sun, 4 Apr 2021 07:35:45 +1000

Subject: [PATCH 2/7] src/pmdas/hacluster/pacemaker.c: plug small mem leak on

 error path in hacluster_refresh_pacemaker_resources() (covscan)

Fixes Coverity CID 366053.

---

 src/pmdas/hacluster/pacemaker.c | 5 ++++-

 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/pmdas/hacluster/pacemaker.c b/src/pmdas/hacluster/pacemaker.c

index 25d80ed41..2dc53f8cf 100644

--- a/src/pmdas/hacluster/pacemaker.c

+++ b/src/pmdas/hacluster/pacemaker.c

@@ -625,8 +625,11 @@ hacluster_refresh_pacemaker_resources(const char *instance_name, struct resource

 	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

-	if ((pf = popen(buffer, "r")) == NULL)

+	if ((pf = popen(buffer, "r")) == NULL) {

+		if (!no_node_attachment)

+		    free(tofree);

 		return -oserror();

+	}

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

-- 

2.31.1

From 625e3a493611864d8785091d0f95a2e1ec293eea Mon Sep 17 00:00:00 2001

From: Paul Evans <pevans@redhat.com>

Date: Fri, 9 Apr 2021 21:10:12 +0100

Subject: [PATCH 3/7] pmdahacluster: Correct logic for

 ha_cluster.pacemaker.resources.* metrics

Correct the logic for pacemaker resources metrics in the hacluster PMDA

simplifying the logic and correcting the issue seen under QA testing

pointed out by Ken along with further corrections in the pacemaker

resources metric collection.

Have also included further buffer overflow guarding for the most of

the sscanf cases in pacemaker.c.

Bonus: Corrected spelling mistakes also pointed out in helptext.

---

 qa/1897.out                     | 32 ++++++++++++++++----------------

 src/pmdas/hacluster/help        | 28 ++++++++++++++--------------

 src/pmdas/hacluster/pacemaker.c | 19 ++++---------------

 3 files changed, 34 insertions(+), 45 deletions(-)

diff --git a/qa/1897.out b/qa/1897.out

index ff9c3ff6c..054b1e92f 100644

--- a/qa/1897.out

+++ b/qa/1897.out

@@ -337,7 +337,7 @@ ha_cluster.pacemaker.nodes.status.dc PMID: 155.3.8 [Whether the node status is g

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the disconnected status is reportered by the node in the cluster, a

+Whether the disconnected status is reported by the node in the cluster, a

 value of 1 confirms the node status as disconnected.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 1

@@ -346,7 +346,7 @@ ha_cluster.pacemaker.nodes.status.expected_up PMID: 155.3.7 [Whether the node st

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the expected_up status is reportered by the node in the cluster, a

+Whether the expected_up status is reported by the node in the cluster, a

 value of 1 confirms the node status as expected_up.

     inst [0 or "node-1"] value 1

     inst [1 or "node-2"] value 1

@@ -355,7 +355,7 @@ ha_cluster.pacemaker.nodes.status.maintenance PMID: 155.3.3 [Whether the node st

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the maintenance status is reportered by the node in the cluster, a

+Whether the maintenance status is reported by the node in the cluster, a

 value of 1 confirms the node status as online.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 0

@@ -364,7 +364,7 @@ ha_cluster.pacemaker.nodes.status.online PMID: 155.3.0 [Whether the node status

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the online status is reportered by the node in the cluster, a value of

+Whether the online status is reported by the node in the cluster, a value of

 1 confirms the node status as online.

     inst [0 or "node-1"] value 1

     inst [1 or "node-2"] value 1

@@ -373,7 +373,7 @@ ha_cluster.pacemaker.nodes.status.pending PMID: 155.3.4 [Whether the node status

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the pending status is reportered by the node in the cluster, a value of

+Whether the pending status is reported by the node in the cluster, a value of

 1 confirms the node status as pending.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 0

@@ -382,7 +382,7 @@ ha_cluster.pacemaker.nodes.status.shutdown PMID: 155.3.6 [Whether the node statu

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the shutdown status is reportered by the node in the cluster, a value

+Whether the shutdown status is reported by the node in the cluster, a value

 of 1 confirms the node status as shutdown.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 0

@@ -391,7 +391,7 @@ ha_cluster.pacemaker.nodes.status.standby PMID: 155.3.1 [Whether the node status

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the standby status is reportered by the node in the cluster, a value of

+Whether the standby status is reported by the node in the cluster, a value of

 1 confirms the node status as standby.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 0

@@ -400,7 +400,7 @@ ha_cluster.pacemaker.nodes.status.standby_on_fail PMID: 155.3.2 [Whether the nod

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the standby_on_fail status is reportered by the node in the cluster,

+Whether the standby_on_fail status is reported by the node in the cluster,

 a value of 1 confirms the node status as standby_on_fail.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 0

@@ -409,7 +409,7 @@ ha_cluster.pacemaker.nodes.status.unclean PMID: 155.3.5 [Whether the node status

     Data Type: 32-bit unsigned int  InDom: 155.2 0x26c00002

     Semantics: instant  Units: count

 Help:

-Whether the unclean status is reportered by the node in the cluster, a value of

+Whether the unclean status is reported by the node in the cluster, a value of

 1 confirms the node status as unclean.

     inst [0 or "node-1"] value 0

     inst [1 or "node-2"] value 0

@@ -466,7 +466,7 @@ ha_cluster.pacemaker.resources.managed PMID: 155.5.3 [Value is either true or fa

     Semantics: instant  Units: count

 Help:

 The value of whether the resource instance in the cluster is managed or not.

-    inst [0 or "my_first_svc:node-1"] value 0

+    inst [0 or "my_first_svc:node-1"] value 1

     inst [5 or "rsc_TEST:node-1"] value 1

     inst [1 or "rsc_Test:node-1"] value 1

     inst [2 or "rsc_Test:node-2"] value 1

@@ -489,9 +489,9 @@ ha_cluster.pacemaker.resources.status.active PMID: 155.5.5 [Whether the resource

     Data Type: 32-bit unsigned int  InDom: 155.4 0x26c00004

     Semantics: instant  Units: count

 Help:

-Whether the active status is reportered by the resource in the cluster, a

+Whether the active status is reported by the resource in the cluster, a

 value of 1 confirms the resource status as active.

-    inst [0 or "my_first_svc:node-1"] value 0

+    inst [0 or "my_first_svc:node-1"] value 1

     inst [5 or "rsc_TEST:node-1"] value 1

     inst [1 or "rsc_Test:node-1"] value 1

     inst [2 or "rsc_Test:node-2"] value 1

@@ -502,7 +502,7 @@ ha_cluster.pacemaker.resources.status.blocked PMID: 155.5.7 [Whether the resourc

     Data Type: 32-bit unsigned int  InDom: 155.4 0x26c00004

     Semantics: instant  Units: count

 Help:

-Whether the blocked status is reportered by the resource in the cluster, a

+Whether the blocked status is reported by the resource in the cluster, a

 value of 1 confirms the resource status as blocked.

     inst [0 or "my_first_svc:node-1"] value 0

     inst [5 or "rsc_TEST:node-1"] value 0

@@ -515,7 +515,7 @@ ha_cluster.pacemaker.resources.status.failed PMID: 155.5.8 [Whether the resource

     Data Type: 32-bit unsigned int  InDom: 155.4 0x26c00004

     Semantics: instant  Units: count

 Help:

-Whether the failed status is reportered by the resource in the cluster, a 

+Whether the failed status is reported by the resource in the cluster, a 

 value of 1 confirms the resource status as failed.

     inst [0 or "my_first_svc:node-1"] value 0

     inst [5 or "rsc_TEST:node-1"] value 0

@@ -528,7 +528,7 @@ ha_cluster.pacemaker.resources.status.failure_ignored PMID: 155.5.9 [Whether the

     Data Type: 32-bit unsigned int  InDom: 155.4 0x26c00004

     Semantics: instant  Units: count

 Help:

-Whether the failure_ignored status is reportered by the resource in the

+Whether the failure_ignored status is reported by the resource in the

 cluster, a value of 1 confirms the resource status as failure_ignored.

     inst [0 or "my_first_svc:node-1"] value 0

     inst [5 or "rsc_TEST:node-1"] value 0

@@ -541,7 +541,7 @@ ha_cluster.pacemaker.resources.status.orphaned PMID: 155.5.6 [Whether the resour

     Data Type: 32-bit unsigned int  InDom: 155.4 0x26c00004

     Semantics: instant  Units: count

 Help:

-Whether the orphaned status is reportered by the resource in the cluster, a

+Whether the orphaned status is reported by the resource in the cluster, a

 value of 1 confirms the resource status as orphaned.

     inst [0 or "my_first_svc:node-1"] value 0

     inst [5 or "rsc_TEST:node-1"] value 0

diff --git a/src/pmdas/hacluster/help b/src/pmdas/hacluster/help

index fa5bf9106..704e940c0 100644

--- a/src/pmdas/hacluster/help

+++ b/src/pmdas/hacluster/help

@@ -64,39 +64,39 @@ Pacemaker internals.

 The membership type given to the node in the Pacemaker cluster.

 @ ha_cluster.pacemaker.nodes.status.online Whether the node status is given as online

-Whether the online status is reportered by the node in the cluster, a value of

+Whether the online status is reported by the node in the cluster, a value of

 1 confirms the node status as online.

 @ ha_cluster.pacemaker.nodes.status.standby Whether the node status is given as standby

-Whether the standby status is reportered by the node in the cluster, a value of

+Whether the standby status is reported by the node in the cluster, a value of

 1 confirms the node status as standby.

 @ ha_cluster.pacemaker.nodes.status.standby_on_fail Whether the node status is given as standby_on_fail

-Whether the standby_on_fail status is reportered by the node in the cluster,

+Whether the standby_on_fail status is reported by the node in the cluster,

 a value of 1 confirms the node status as standby_on_fail.

 @ ha_cluster.pacemaker.nodes.status.maintenance Whether the node status is given as maintenance

-Whether the maintenance status is reportered by the node in the cluster, a

+Whether the maintenance status is reported by the node in the cluster, a

 value of 1 confirms the node status as online.

 @ ha_cluster.pacemaker.nodes.status.pending Whether the node status is given as pending

-Whether the pending status is reportered by the node in the cluster, a value of

+Whether the pending status is reported by the node in the cluster, a value of

 1 confirms the node status as pending.

 @ ha_cluster.pacemaker.nodes.status.unclean Whether the node status is given as unclean

-Whether the unclean status is reportered by the node in the cluster, a value of

+Whether the unclean status is reported by the node in the cluster, a value of

 1 confirms the node status as unclean.

 @ ha_cluster.pacemaker.nodes.status.shutdown Whether the node status is given as shutdown

-Whether the shutdown status is reportered by the node in the cluster, a value

+Whether the shutdown status is reported by the node in the cluster, a value

 of 1 confirms the node status as shutdown.

 @ ha_cluster.pacemaker.nodes.status.expected_up Whether the node status is given as expected_up

-Whether the expected_up status is reportered by the node in the cluster, a

+Whether the expected_up status is reported by the node in the cluster, a

 value of 1 confirms the node status as expected_up.

 @ ha_cluster.pacemaker.nodes.status.dc Whether the node status is given as disconnected

-Whether the disconnected status is reportered by the node in the cluster, a

+Whether the disconnected status is reported by the node in the cluster, a

 value of 1 confirms the node status as disconnected.

 @ ha_cluster.pacemaker.resources.agent The name of the resource agent for this resource

@@ -118,23 +118,23 @@ The value of whether the resource instance in the cluster is managed or not.

 The value of the given role state for the resource instance in the cluster.

 @ ha_cluster.pacemaker.resources.status.active Whether the resources status is given as active

-Whether the active status is reportered by the resource in the cluster, a

+Whether the active status is reported by the resource in the cluster, a

 value of 1 confirms the resource status as active.

 @ ha_cluster.pacemaker.resources.status.orphaned Whether the resources status is given as orphaned

-Whether the orphaned status is reportered by the resource in the cluster, a

+Whether the orphaned status is reported by the resource in the cluster, a

 value of 1 confirms the resource status as orphaned.

 @ ha_cluster.pacemaker.resources.status.blocked Whether the resources status is given as blocked

-Whether the blocked status is reportered by the resource in the cluster, a

+Whether the blocked status is reported by the resource in the cluster, a

 value of 1 confirms the resource status as blocked.

 @ ha_cluster.pacemaker.resources.status.failed Whether the resources status is given as failed

-Whether the failed status is reportered by the resource in the cluster, a 

+Whether the failed status is reported by the resource in the cluster, a 

 value of 1 confirms the resource status as failed.

 @ ha_cluster.pacemaker.resources.status.failure_ignored Whether the resources status is given as failure_ignored

-Whether the failure_ignored status is reportered by the resource in the

+Whether the failure_ignored status is reported by the resource in the

 cluster, a value of 1 confirms the resource status as failure_ignored.

 @ ha_cluster.corosync.quorate Value given for quorate

diff --git a/src/pmdas/hacluster/pacemaker.c b/src/pmdas/hacluster/pacemaker.c

index 2dc53f8cf..8f8e626a5 100644

--- a/src/pmdas/hacluster/pacemaker.c

+++ b/src/pmdas/hacluster/pacemaker.c

@@ -513,7 +513,7 @@ hacluster_refresh_pacemaker_nodes(const char *node_name, struct nodes *nodes)

 		/* Collect our node names */

 		if (found_nodes && strstr(buffer, node_name)) {

-			sscanf(buffer, "%*s %*s %*s online=\"%[^\"]\" standby=\"%[^\"]\" standby_onfail=\"%[^\"]\" maintenance=\"%[^\"]\" pending=\"%[^\"]\" unclean=\"%[^\"]\" shutdown=\"%[^\"]\" expected_up=\"%[^\"]\" is_dc =\"%[^\"]\" %*s type=\"%[^\"]\"",

+			sscanf(buffer, "%*s %*s %*s online=\"%9[^\"]\" standby=\"%9[^\"]\" standby_onfail=\"%9[^\"]\" maintenance=\"%9[^\"]\" pending=\"%9[^\"]\" unclean=\"%9[^\"]\" shutdown=\"%9[^\"]\" expected_up=\"%9[^\"]\" is_dc =\"%9[^\"]\" %*s type=\"%9[^\"]\"",

 				online,

 				standby,

 				standby_on_fail,

@@ -671,19 +671,8 @@ hacluster_refresh_pacemaker_resources(const char *instance_name, struct resource

 			/* Collect our metrics */

 			if (strstr(buffer, "resource id=") && strstr(buffer, resource_id)) {

-				if(strstr(resources->clone, "\0") || strstr(resources->group, "\0")) {

-					sscanf(buffer, "%*s %*s resource_agent=\"%[^\"]\" role=\"%[^\"]\" active=\"%[^\"]\" orphaned=\"%[^\"]\" blocked=\"%[^\"]\" managed=\"%[^\"]\" failed=\"%[^\"]\" failure_ignored=\"%[^\"]\"",

-						resources->agent,

-						resources->role,

-						active,

-						orphaned,

-						blocked,

-						managed,

-						failed,

-						failure_ignored

-					);

-				} else if ((strstr(resources->clone, "\0") || strstr(resources->group, "\0")) && strstr(buffer, "target_role")) { 

-					sscanf(buffer, "%*s %*s resource_agent=\"%[^\"]\" role=\"%[^\"]\" %*s active=\"%[^\"]\" orphaned=\"%[^\"]\" blocked=\"%[^\"]\" managed=\"%[^\"]\" failed=\"%[^\"]\" failure_ignored=\"%[^\"]\"",

+				if (strstr(buffer, "target_role")) { 

+					sscanf(buffer, "%*s %*s resource_agent=\"%[^\"]\" role=\"%[^\"]\" %*s active=\"%7[^\"]\" orphaned=\"%7[^\"]\" blocked=\"%7[^\"]\" managed=\"%7[^\"]\" failed=\"%7[^\"]\" failure_ignored=\"%7[^\"]\"",

 						resources->agent,

 						resources->role,

 						active,

@@ -694,7 +683,7 @@ hacluster_refresh_pacemaker_resources(const char *instance_name, struct resource

 						failure_ignored

 					);

 				} else {

-					sscanf(buffer, "%*s %*s resource_agent=\"%[^\"]\" role=\"%[^\"]\" %*s active=\"%[^\"]\" orphaned=\"%[^\"]\" blocked=\"%[^\"]\" managed=\"%[^\"]\" failed=\"%[^\"]\" failure_ignored=\"%[^\"]\"",

+					sscanf(buffer, "%*s %*s resource_agent=\"%[^\"]\" role=\"%[^\"]\" active=\"%7[^\"]\" orphaned=\"%7[^\"]\" blocked=\"%7[^\"]\" managed=\"%7[^\"]\" failed=\"%7[^\"]\" failure_ignored=\"%7[^\"]\"",

 						resources->agent,

 						resources->role,

 						active,

-- 

2.31.1

From 7f39eb78a86b244a046d7014c744abe21b3bef52 Mon Sep 17 00:00:00 2001

From: Paul Evans <pevans@redhat.com>

Date: Fri, 9 Apr 2021 21:12:19 +0100

Subject: [PATCH 4/7] pmdahacluster: Reduce log messages from popen()

The PMDA is designed to be able to collect metrics on setups where not all

of the HA Cluster components might exist. Suppress some of these logging

warnings when certain components are not available on the host system.

---

 src/pmdas/hacluster/corosync.c  | 20 +++++++--------

 src/pmdas/hacluster/drbd.c      |  8 +++---

 src/pmdas/hacluster/pacemaker.c | 14 +++++------

 src/pmdas/hacluster/pmda.c      | 44 +++++++++++++++++++--------------

 src/pmdas/hacluster/sbd.c       |  4 +--

 5 files changed, 49 insertions(+), 41 deletions(-)

diff --git a/src/pmdas/hacluster/corosync.c b/src/pmdas/hacluster/corosync.c

index c7964b621..ee0052dee 100644

--- a/src/pmdas/hacluster/corosync.c

+++ b/src/pmdas/hacluster/corosync.c

@@ -139,10 +139,10 @@ hacluster_refresh_corosync_node(const char *node_name, struct member_votes *node

 	char *buffer_ptr;

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s", quorumtool_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", quorumtool_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		if (strstr(buffer, node_name)) {

@@ -187,10 +187,10 @@ hacluster_refresh_corosync_global()

 	char buffer[4096], quorate[6];

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s", quorumtool_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", quorumtool_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

@@ -218,10 +218,10 @@ hacluster_refresh_corosync_global()

 	}

 	pclose(pf);

-	pmsprintf(buffer, sizeof(buffer), "%s", cfgtool_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", cfgtool_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

@@ -240,10 +240,10 @@ hacluster_refresh_corosync_ring(const char *ring_name, struct rings *rings)

 	FILE *pf;	

 	int ring_found = 0;

-	pmsprintf(buffer, sizeof(buffer), "%s", cfgtool_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", cfgtool_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

@@ -293,10 +293,10 @@ hacluster_refresh_corosync_ring(const char *ring_name, struct rings *rings)

 	}

 	pclose(pf);

-	pmsprintf(buffer, sizeof(buffer), "%s", quorumtool_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", quorumtool_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();  

+		return oserror();

 	/* 

 	 * Check corosync-quorumtool for our node_id and ring_id values for our

diff --git a/src/pmdas/hacluster/drbd.c b/src/pmdas/hacluster/drbd.c

index 7fb3b04d0..bec83031c 100644

--- a/src/pmdas/hacluster/drbd.c

+++ b/src/pmdas/hacluster/drbd.c

@@ -156,10 +156,10 @@ hacluster_refresh_drbd_resource(const char *resource_name, struct resource *reso

 	int found_node = 0, found_volume = 0, nesting = 0;

-	pmsprintf(buffer, sizeof(buffer), "%s", drbdsetup_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", drbdsetup_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	/* 

 	 * We need to split our combined NODE:VOLUME instance names into their

@@ -274,10 +274,10 @@ hacluster_refresh_drbd_peer_device(const char *peer_name, struct peer_device *pe

 	int found_node = 0, found_peer_node = 0, nesting = 0;

-	pmsprintf(buffer, sizeof(buffer), "%s", drbdsetup_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", drbdsetup_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	/* 

 	 * We need to split our combined NODE:PEER_NODE_ID instance names into

diff --git a/src/pmdas/hacluster/pacemaker.c b/src/pmdas/hacluster/pacemaker.c

index 8f8e626a5..355f4df5c 100644

--- a/src/pmdas/hacluster/pacemaker.c

+++ b/src/pmdas/hacluster/pacemaker.c

@@ -360,7 +360,7 @@ hacluster_refresh_pacemaker_global()

 	char last_written_text[128], stonith[6];

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s", cibadmin_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", cibadmin_command);

 	if ((pf = popen(buffer, "r")) == NULL)

 		return -oserror();

@@ -374,7 +374,7 @@ hacluster_refresh_pacemaker_global()

 	}

 	pclose(pf);

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

 		return -oserror();

@@ -402,7 +402,7 @@ hacluster_refresh_pacemaker_fail(const char *instance_name, struct fail_count *f

 	int found_node_history = 0, found_node_name = 0;

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

 		return -oserror();

@@ -455,7 +455,7 @@ hacluster_refresh_pacemaker_constraints(const char *constraints_name, struct loc

 	int found_constraints = 0;

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s", cibadmin_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", cibadmin_command);

 	if ((pf = popen(buffer, "r")) == NULL)

 		return -oserror();

@@ -492,7 +492,7 @@ hacluster_refresh_pacemaker_nodes(const char *node_name, struct nodes *nodes)

 	char online[10], standby[10], standby_on_fail[10], maintenance[10], pending[10];

 	char unclean[10], shutdown[10], expected_up[10], dc[10];

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

 		return -oserror();

@@ -549,7 +549,7 @@ hacluster_refresh_pacemaker_node_attribs(const char *attrib_name, struct attribu

 	int found_node_attributes = 0, found_node_name = 0;

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

 		return -oserror();

@@ -623,7 +623,7 @@ hacluster_refresh_pacemaker_resources(const char *instance_name, struct resource

 		node = strsep(&str, ":");

 	}

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL) {

 		if (!no_node_attachment)

diff --git a/src/pmdas/hacluster/pmda.c b/src/pmdas/hacluster/pmda.c

index 196de2e64..c7ffac6da 100644

--- a/src/pmdas/hacluster/pmda.c

+++ b/src/pmdas/hacluster/pmda.c

@@ -359,10 +359,10 @@ hacluster_pacemaker_fail_instance_refresh(void)

 	FILE		*pf;

 	pmInDom		indom = INDOM(PACEMAKER_FAIL_INDOM);

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* First we need to check whether we are in <node_history> section*/

@@ -426,11 +426,11 @@ hacluster_pacemaker_constraints_instance_refresh(void)

 	FILE		*pf;

 	pmInDom		indom = INDOM(PACEMAKER_CONSTRAINTS_INDOM);

-	pmsprintf(buffer, sizeof(buffer), "%s", cibadmin_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", cibadmin_command);

 	buffer[sizeof(buffer)-1] = '\0';

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* First we need to check whether we are in <constraints> section*/

@@ -472,10 +472,10 @@ hacluster_pacemaker_nodes_instance_refresh(void)

 	FILE		*pf;

 	pmInDom		indom = INDOM(PACEMAKER_NODES_INDOM);

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* First we need to check whether we are in <nodes> section*/

@@ -525,10 +525,10 @@ hacluster_pacemaker_node_attrib_instance_refresh(void)

 	FILE		*pf;

 	pmInDom		indom = INDOM(PACEMAKER_NODE_ATTRIB_INDOM);

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* First we need to check whether we are in <node_history> section*/

@@ -598,10 +598,10 @@ hacluster_pacemaker_resources_instance_refresh(void)

 	FILE		*pf;

 	pmInDom		indom= INDOM(PACEMAKER_RESOURCES_INDOM);

-	pmsprintf(buffer, sizeof(buffer), "%s", crm_mon_command);

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", crm_mon_command);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* First we need to check whether we are in <resources> section*/

@@ -676,9 +676,11 @@ hacluster_corosync_node_instance_refresh(void)

 	 * membership information section of corosync-quorumtool output

 	 */

 	pmdaCacheOp(indom, PMDA_CACHE_INACTIVE);

+	

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", quorumtool_command);

-	if ((pf = popen(quorumtool_command, "r")) == NULL)

-		return -oserror();

+	if ((pf = popen(buffer, "r")) == NULL)

+		return oserror();

 	while (fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* Clear whitespace at start of each line */

@@ -735,8 +737,10 @@ hacluster_corosync_ring_instance_refresh(void)

 	 */

 	pmdaCacheOp(indom, PMDA_CACHE_INACTIVE);

-	if ((pf = popen(cfgtool_command, "r")) == NULL)

-		return -oserror();

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", cfgtool_command);

+	

+	if ((pf = popen(buffer, "r")) == NULL)

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

@@ -865,8 +869,10 @@ hacluster_drbd_resource_instance_refresh(void)

 	 */

 	pmdaCacheOp(indom, PMDA_CACHE_INACTIVE);

-	if ((pf = popen(drbdsetup_command, "r")) == NULL)

-		return -oserror();

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", drbdsetup_command);

+	

+	if ((pf = popen(buffer, "r")) == NULL)

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* Clear whitespace at start of each line */

@@ -940,8 +946,10 @@ hacluster_drbd_peer_device_instance_refresh(void)

 	 */

 	pmdaCacheOp(indom, PMDA_CACHE_INACTIVE);

-	if ((pf = popen(drbdsetup_command, "r")) == NULL)

-		return -oserror();

+	pmsprintf(buffer, sizeof(buffer), "%s 2>&1", drbdsetup_command);

+

+	if ((pf = popen(buffer, "r")) == NULL)

+		return oserror();

 	while(fgets(buffer, sizeof(buffer)-1, pf) != NULL) {

 		/* Clear whitespace at start of each line */

diff --git a/src/pmdas/hacluster/sbd.c b/src/pmdas/hacluster/sbd.c

index 5f55d0734..2824f6589 100644

--- a/src/pmdas/hacluster/sbd.c

+++ b/src/pmdas/hacluster/sbd.c

@@ -74,10 +74,10 @@ hacluster_refresh_sbd_device(const char *sbd_dev, struct sbd *sbd)

 	char buffer[4096];

 	FILE *pf;

-	pmsprintf(buffer, sizeof(buffer), "%s -d %s dump", sbd_command, sbd_dev);

+	pmsprintf(buffer, sizeof(buffer), "%s -d %s dump 2>&1", sbd_command, sbd_dev);

 	if ((pf = popen(buffer, "r")) == NULL)

-		return -oserror();

+		return oserror();

 	strncpy(sbd->path, sbd_dev, sizeof(sbd->path));

 	sbd->path[sizeof(sbd->path)-1] = '\0';

-- 

2.31.1

From 91241a5f76fc83895621fe35f399b5dcd97e796e Mon Sep 17 00:00:00 2001

From: Paul Evans <pevans@redhat.com>

Date: Fri, 18 Jun 2021 00:30:39 +0100

Subject: [PATCH 5/7] pmdahacluster: Add support for labels on key metrics

Introduction of a number of labels on some metrics allowing the export

of metadata in cases where PM_TYPE_STRING is not supported.

This will allow information including Pacemaker location constraint data and

others to be successfully exported in pmproxy using the OpenMetrics API.

Satisfies RHBZ# 1972277: PCP - PMDA HA Cluster (pcp-pmda-hacluster) - unable

to export pacemaker location constraints data due to unsupported metric type

---

 qa/1897.out                         |  66 ++++-

 src/pmdas/hacluster/corosync.c      |   7 +

 src/pmdas/hacluster/corosync.h      |   1 +

 src/pmdas/hacluster/drbd.c          |  14 ++

 src/pmdas/hacluster/drbd.h          |   2 +

 src/pmdas/hacluster/help            |  30 ++-

 src/pmdas/hacluster/pacemaker.c     |  21 ++

 src/pmdas/hacluster/pacemaker.h     |   3 +

 src/pmdas/hacluster/pmda.c          | 369 ++++++++++++++++++++++++++--

 src/pmdas/hacluster/pmdahacluster.h |  58 +++--

 src/pmdas/hacluster/pmns            | 119 ++++-----

 src/pmdas/hacluster/sbd.c           |   7 +

 src/pmdas/hacluster/sbd.h           |   1 +

 13 files changed, 602 insertions(+), 96 deletions(-)

diff --git a/qa/1897.out b/qa/1897.out

index 054b1e92f..b757e857a 100644

--- a/qa/1897.out