File 0001-TimeoutFormulas.patch of Package crmsh
From 2b40cbd002bea83ba29f867993fb8ec22d41c41a Mon Sep 17 00:00:00 2001 From: xin liang <xliang@suse.com> Date: Tue, 29 Jul 2025 16:47:13 +0800 Subject: [PATCH] Fix: doc: Add TimeoutFormulas help topic (bsc#1242981) To show SBD and Pacemaker related timeout formulas in the help topic. --- doc/crm.8.adoc | 67 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 67 insertions(+) diff --git a/doc/crm.8.adoc b/doc/crm.8.adoc index 9d9a9111..ecdf8f1e 100644 --- a/doc/crm.8.adoc +++ b/doc/crm.8.adoc @@ -296,6 +296,73 @@ INFO: test-2 shadow CIB created crm(test-2)configure# commit ............... +[[topics.TimeoutFormulas,Timeout formulas]] +=== Timeout formulas + +If you use `crm cluster init` command to set up a cluster with SBD as +the stonith device, crmsh configures several fencing related settings +based on the calculations with a set of formulas. + +.SBD related timeouts +`watchdog` timeout: Default value is 15s. This value is from the definition +in +/etc/crm/profiles.xml+, can be a different value with a different profile. +`msgwait` timeout: +............... + msgwait = 2 * watchdog +............... +`SBD_WATCHDOG_TIMEOUT`: +............... + # For diskless SBD + SBD_WATCHDOG_TIMEOUT = Default value is 15s. This value is from the definition +in +/etc/crm/profiles.xml+, can be a different value with a different profile. + + # When adding diskless SBD and QDevice together from crm cluster init + if SBD_WATCHDOG_TIMEOUT < 35s: + SBD_WATCHDOG_TIMEOUT = 35s + + # When adding diskless SBD in a cluster with QDevice already + if SBD_WATCHDOG_TIMEOUT < qdevice_sync_timeout: + SBD_WATCHDOG_TIMEOUT = qdevice_sync_timeout + 5s + + # When adding QDevice in a cluster with diskless SBD already + if SBD_WATCHDOG_TIMEOUT < 35s: + SBD_WATCHDOG_TIMEOUT = 35s +............... +`SBD_DELAY_START`: +............... + # For disk-based SBD + SBD_DELAY_START = pcmk_delay_max + msgwait + corosync.token + corosync.consensus + + # For diskless SBD + SBD_DELAY_START = 2 * SBD_WATCHDOG_TIMEOUT + corosync.token + corosync.consensus +............... +`TimeoutSec` of sbd.service systemd unit: +............... + default_value = 1min30s + if SBD_DELAY_START > default_value: + start_timeout = 1.2 * SBD_DELAY_START + else: + start_timeout = default_value +............... +.Pacemaker related timeouts +`stonith-watchdog-timeout`: (for diskless SBD) +............... + stonith-watchdog-timeout = 2 * SBD_WATCHDOG_TIMEOUT +............... +`stonith-timeout`: +............... + # For disk-based SBD + value_from_sbd = 1.2 * msgwait + stonith-timeout = max(value_from_sbd, 60) + corosync.token + corosync.consensus + + # For diskless SBD + value_from_sbd = 1.2 * max(stonith-watchdog-timeout, 2*SBD_WATCHDOG_TIMEOUT) + stonith-timeout = max(value_from_sbd, 60) + corosync.token + corosync.consensus + + # For general scenarios + stonith-timeout = 60 + corosync.token + corosync.consensus +............... + [[topics.Features.Checks,Configuration semantic checks]] === Configuration semantic checks -- 2.45.2
