From 156a0ca201ce227864ea38e249d3e5094c70bb3e Mon Sep 17 00:00:00 2001
From: digimer <mkelly@alteeve.ca>
Date: Fri, 16 Jun 2023 11:43:49 -0400
Subject: [PATCH] Updated anvil-daemon's new job launching logic to allow the
 restart of a running job that failed out early.

Signed-off-by: digimer <mkelly@alteeve.ca>
---
 Anvil/Tools.pm          |  1 +
 share/words.xml         |  1 +
 tools/anvil-daemon      | 17 ++++++++++++++++-
 tools/anvil-watch-power |  2 ++
 4 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/Anvil/Tools.pm b/Anvil/Tools.pm
index fcd00b3a..4cd64e43 100644
--- a/Anvil/Tools.pm
+++ b/Anvil/Tools.pm
@@ -1095,6 +1095,7 @@ sub _set_paths
 				bonds				=>	"/proc/net/bonding",
 				'cgi-bin'			=>	"/var/www/cgi-bin",
 				drbd_resources			=>	"/etc/drbd.d/",
+				drbd_kernel_proc		=>	"/sys/kernel/debug/drbd/resources",
 				fence_agents			=>	"/usr/sbin",
 				firewalld_services		=>	"/usr/lib/firewalld/services",
 				firewalld_zones_etc		=>	"/etc/firewalld/zones",		# Changes when firewall-cmd ... --permanent is used.
diff --git a/share/words.xml b/share/words.xml
index 6aa50e92..f61139c6 100644
--- a/share/words.xml
+++ b/share/words.xml
@@ -2393,6 +2393,7 @@ The file: [#!variable!file!#] needs to be updated. The difference is:
 		<key name="log_0739">The server: [#!variable!server!#] needs it's pacemaker configuration updated. Running: [#!variable!command!#].</key>
 		<key name="log_0740">Running the scan-agent: [#!variable!agent!#] now to ensure that the database has an updated view of resources.</key>
 		<key name="log_0741">I was about to start: [#!variable!command!#] with the job UUID: [#!variable!this_job_uuid!#]. However, another job using the same command with the job UUID: [#!variable!other_job_uuid!#]. To avoid race conditions, only one process with a given command is run at the same time.</key>
+		<key name="log_0742">The job with the command: [#!variable!command!#] and job UUID: [#!variable!job_uuid!#] is restarting.</key>
 
 		<!-- Messages for users (less technical than log entries), though sometimes used for logs, too. -->
 		<key name="message_0001">The host name: [#!variable!target!#] does not resolve to an IP address.</key>
diff --git a/tools/anvil-daemon b/tools/anvil-daemon
index 3c1be196..12f167f8 100755
--- a/tools/anvil-daemon
+++ b/tools/anvil-daemon
@@ -1637,7 +1637,22 @@ sub run_jobs
 					
 					if (exists $anvil->data->{jobs}{running}{$started_job_uuid})
 					{
-						if ($anvil->data->{jobs}{running}{$started_job_uuid}{job_progress} != 100)
+						# If the previously running job and this job have the same
+						# UUID, it failed and needs to restart.
+						$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { 
+							job_uuid                                           => $job_uuid, 
+							started_job_uuid                                   => $started_job_uuid,
+							"jobs::running::${started_job_uuid}::job_progress" => $anvil->data->{jobs}{running}{$started_job_uuid}{job_progress}, 
+						}});
+						if ($started_job_uuid eq $job_uuid)
+						{
+							# We're restarting.
+							$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0741", variables => { 
+								command  => $job_command,
+								job_uuid => $job_uuid, 
+							}});
+						}
+						elsif ($anvil->data->{jobs}{running}{$started_job_uuid}{job_progress} != 100)
 						{
 							# Don't start it in this pass.
 							$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0741", variables => { 
diff --git a/tools/anvil-watch-power b/tools/anvil-watch-power
index 197f74ca..20dc7af5 100755
--- a/tools/anvil-watch-power
+++ b/tools/anvil-watch-power
@@ -52,6 +52,8 @@ while(1)
 	sleep 2;
 }
 
+$anvil->nice_exit({exit_code => 0});
+
 
 sub show_power_data
 {