File 0002-crm-fence-peer-fix-timeout-with-Pacemaker-... of Package drbd-utils

Overview Repositories Revisions Requests Users Attributes Meta

File 0002-crm-fence-peer-fix-timeout-with-Pacemaker-2.0.5-mill.patch of Package drbd-utils (Revision 743b3162c11290af827d7f76549f2326)

Currently displaying revision 743b3162c11290af827d7f76549f2326 , Show latest

From 68d1e4242f165917bc2c787d9df0fe41251e05e6 Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Wed, 12 Jan 2022 13:50:35 +0100
Subject: [PATCH] crm-fence-peer: fix timeout with Pacemaker 2.0.5: milli
 seconds vs seconds
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Addendum to 8a28be74bc6efa93931c957e54c01abb18b984fe
Commit message of the above cited here:

> crmadmin timeout was in milli seconds for <= 2.0.x,
> but became a TIMESPEC with default seconds in >= 2.1.
>
> Up to 2.0.4, atoi() was used, which effectively ignores "trailing garbage",
> so we could get away with always appending "ms".
> But with 2.0.5, it became g_option_context_parse G_OPTION_ARG_INT, which
> "Cannot parse integer value “200ms” for --timeout" :-|
>
> So grep the help message for "timeout.*milliseconds",
> and if not present, append an explicit "ms" unit.

And this is where I got it wrong :-(
somewhere later they re-organised the help text
so now I would need to parse --help-all.

Instead try to actually call "crmadmin -t 100ms --version".
If that works, it apparently understands (or ignores)
the "ms" unit.
---
 scripts/crm-fence-peer.9.sh | 14 +++++++++++---
 scripts/crm-fence-peer.sh   | 14 +++++++++++---
 2 files changed, 22 insertions(+), 6 deletions(-)

diff --git a/scripts/crm-fence-peer.9.sh b/scripts/crm-fence-peer.9.sh
index c943bf9f..fc8d2bc3 100755
--- a/scripts/crm-fence-peer.9.sh
+++ b/scripts/crm-fence-peer.9.sh
@@ -400,10 +400,18 @@ setup_crm_timeout_unit_ms()
 	# garbage", so we could get away with always appending "ms", but with
 	# 2.0.5, it became g_option_context_parse G_OPTION_ARG_INT, which
 	# "Cannot parse integer value “200ms” for --timeout" :-|
-	if crmadmin --help 2>&1 | grep -q -e "--timeout=.*in milliseconds"; then
-		crm_timeout_unit_ms=""
-	else
+	# Can not parse the help text reliably, because they changed content
+	# and organisation of the help text between 2.0.4 and 2.0.5.
+	# Just try using ms unit, and see if it fails.
+	if crmadmin -t 100ms --version &> /dev/null; then
+		# this is either a recent version that actually understands ms
+		# as part of the TIMESPEC, or a version that still uses atoi().
 		crm_timeout_unit_ms="ms"
+	else
+		# this one likely failed with
+		# crmadmin: Cannot parse integer value “100ms” for -t
+		# (>= 2.0.5, < 2.1)
+		crm_timeout_unit_ms=""
 	fi
 }
 
diff --git a/scripts/crm-fence-peer.sh b/scripts/crm-fence-peer.sh
index 96786734..b0e4e0f1 100755
--- a/scripts/crm-fence-peer.sh
+++ b/scripts/crm-fence-peer.sh
@@ -252,10 +252,18 @@ setup_crm_timeout_unit_ms()
 	# garbage", so we could get away with always appending "ms", but with
 	# 2.0.5, it became g_option_context_parse G_OPTION_ARG_INT, which
 	# "Cannot parse integer value “200ms” for --timeout" :-|
-	if crmadmin --help 2>&1 | grep -q -e "--timeout=.*in milliseconds"; then
-		crm_timeout_unit_ms=""
-	else
+	# Can not parse the help text reliably, because they changed content
+	# and organisation of the help text between 2.0.4 and 2.0.5.
+	# Just try using ms unit, and see if it fails.
+	if crmadmin -t 100ms --version &> /dev/null; then
+		# this is either a recent version that actually understands ms
+		# as part of the TIMESPEC, or a version that still uses atoi().
 		crm_timeout_unit_ms="ms"
+	else
+		# this one likely failed with
+		# crmadmin: Cannot parse integer value “100ms” for -t
+		# (>= 2.0.5, < 2.1)
+		crm_timeout_unit_ms=""
 	fi
 }
 
-- 
2.40.0

Places

File 0002-crm-fence-peer-fix-timeout-with-Pacemaker-2.0.5-mill.patch of Package drbd-utils (Revision 743b3162c11290af827d7f76549f2326)

Places