From 11a63822fbdc0a9ebe1b668b26a59f1cc9649f6c Mon Sep 17 00:00:00 2001

From: Oyvind Albrigtsen <oalbrigt@redhat.com>

Date: Wed, 24 Oct 2018 14:51:27 +0200

Subject: [PATCH] fence_scsi: watchdog retries support

---

 agents/scsi/fence_scsi.py          | 60 ++++++++++++++++++++----------

 tests/data/metadata/fence_scsi.xml |  4 +-

 2 files changed, 43 insertions(+), 21 deletions(-)

diff --git a/agents/scsi/fence_scsi.py b/agents/scsi/fence_scsi.py

index 79ada4fa..8a1e4c77 100644

--- a/agents/scsi/fence_scsi.py

+++ b/agents/scsi/fence_scsi.py

@@ -158,13 +158,15 @@ def get_reservation_key(options, dev):

 	return match.group(1) if match else None

-def get_registration_keys(options, dev):

+def get_registration_keys(options, dev, fail=True):

 	reset_dev(options,dev)

 	keys = []

 	cmd = options["--sg_persist-path"] + " -n -i -k -d " + dev

 	out = run_cmd(options, cmd)

 	if out["err"]:

-		fail_usage("Cannot get registration keys")

+		fail_usage("Cannot get registration keys", fail)

+		if not fail:

+			return []

 	for line in out["out"].split("\n"):

 		match = re.search(r"\s+0x(\S+)\s*", line)

 		if match:

@@ -218,9 +220,8 @@ def get_key(fail=True):

 	try:

 		f = open(file_path, "r")

 	except IOError:

-		if fail:

-			fail_usage("Failed: Cannot open file \""+ file_path + "\"")

-		else:

+		fail_usage("Failed: Cannot open file \""+ file_path + "\"", fail)

+		if not fail:

 			return None

 	return f.readline().strip().lower()

@@ -244,9 +245,8 @@ def dev_read(fail=True):

 	try:

 		f = open(file_path, "r")

 	except IOError:

-		if fail:

-			fail_usage("Failed: Cannot open file \"" + file_path + "\"")

-		else:

+		fail_usage("Failed: Cannot open file \"" + file_path + "\"", fail)

+		if not fail:

 			return None

 	# get not empty lines from file

 	devs = [line.strip() for line in f if line.strip()]

@@ -371,14 +371,20 @@ def define_new_opts():

 	}

-def scsi_check_get_verbose():

+def scsi_check_get_options(options):

 	try:

-		f = open("/etc/sysconfig/watchdog", "r")

+		f = open("/etc/sysconfig/stonith", "r")

 	except IOError:

-		return False

-	match = re.search(r"^\s*verbose=yes", "".join(f.readlines()), re.MULTILINE)

+		return options

+

+	match = re.findall(r"^\s*(\S*)\s*=\s*(\S*)\s*", "".join(f.readlines()), re.MULTILINE)

+

+	for m in match:

+		options[m[0].lower()] = m[1].lower()

+

 	f.close()

-	return bool(match)

+

+	return options

 def scsi_check(hardreboot=False):

@@ -388,7 +394,10 @@ def scsi_check(hardreboot=False):

 	options["--sg_turs-path"] = "@SG_TURS_PATH@"

 	options["--sg_persist-path"] = "@SG_PERSIST_PATH@"

 	options["--power-timeout"] = "5"

-	if scsi_check_get_verbose():

+	options["retry"] = "0"

+	options["retry-sleep"] = "1"

+	options = scsi_check_get_options(options)

+	if "verbose" in options and options["verbose"] == "yes":

 		logging.getLogger().setLevel(logging.DEBUG)

 	devs = dev_read(fail=False)

 	if not devs:

@@ -399,11 +408,18 @@ def scsi_check(hardreboot=False):

 		logging.error("Key not found")

 		return 0

 	for dev in devs:

-		if key in get_registration_keys(options, dev):

-			logging.debug("key " + key + " registered with device " + dev)

-			return 0

-		else:

-			logging.debug("key " + key + " not registered with device " + dev)

+		for n in range(int(options["retry"]) + 1):

+			if n > 0:

+				logging.debug("retry: " + str(n) + " of " + options["retry"])

+			if key in get_registration_keys(options, dev, fail=False):

+				logging.debug("key " + key + " registered with device " + dev)

+				return 0

+			else:

+				logging.debug("key " + key + " not registered with device " + dev)

+

+			if n < int(options["retry"]):

+				time.sleep(float(options["retry-sleep"]))

+

 	logging.debug("key " + key + " registered with any devices")

 	if hardreboot == True:

@@ -452,7 +468,11 @@ def main():

 device(s). The result is that only registered nodes may write to the \

 device(s). When a node failure occurs, the fence_scsi agent will remove the \

 key belonging to the failed node from the device(s). The failed node will no \

-longer be able to write to the device(s). A manual reboot is required."

+longer be able to write to the device(s). A manual reboot is required.\

+\n.P\n\

+When used as a watchdog device you can define e.g. retry=1, retry-sleep=2 and \

+verbose=yes parameters in /etc/sysconfig/stonith if you have issues with it \

+failing."

 	docs["vendorurl"] = ""

 	show_docs(options, docs)

diff --git a/tests/data/metadata/fence_scsi.xml b/tests/data/metadata/fence_scsi.xml

index 45a84168..b8cdabd1 100644

--- a/tests/data/metadata/fence_scsi.xml

+++ b/tests/data/metadata/fence_scsi.xml

@@ -1,7 +1,9 @@

 <resource-agent name="fence_scsi" shortdesc="Fence agent for SCSI persistent reservation" >

 <longdesc>fence_scsi is an I/O fencing agent that uses SCSI-3 persistent reservations to control access to shared storage devices. These devices must support SCSI-3 persistent reservations (SPC-3 or greater) as well as the "preempt-and-abort" subcommand.

-The fence_scsi agent works by having each node in the cluster register a unique key with the SCSI device(s). Once registered, a single node will become the reservation holder by creating a "write exclusive, registrants only" reservation on the device(s). The result is that only registered nodes may write to the device(s). When a node failure occurs, the fence_scsi agent will remove the key belonging to the failed node from the device(s). The failed node will no longer be able to write to the device(s). A manual reboot is required.</longdesc>

+The fence_scsi agent works by having each node in the cluster register a unique key with the SCSI device(s). Once registered, a single node will become the reservation holder by creating a "write exclusive, registrants only" reservation on the device(s). The result is that only registered nodes may write to the device(s). When a node failure occurs, the fence_scsi agent will remove the key belonging to the failed node from the device(s). The failed node will no longer be able to write to the device(s). A manual reboot is required.

+

+When used as a watchdog device you can define e.g. retry=1, retry-sleep=2 and verbose=yes parameters in /etc/sysconfig/stonith if you have issues with it failing.</longdesc>

 <vendor-url></vendor-url>

 <parameters>

 	<parameter name="action" unique="0" required="1">