Local modifications to ClusterLabs/Anvil by Alteeve
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1384 lines
52 KiB

#!/usr/bin/perl
#
# This is the master daemon that manages all periodically run processes on Striker dashboards, Anvil! cluster
# nodes and DR hosts.
#
# Exit codes;
# 0 = Normal exit or md5sum of this program changed and it exited to reload.
# 1 = Not running as root.
# 2 = Unable to connect to any database, even after trying to initialize the local system.
#
# TODO:
# - Need to check what kind of machine this is and not prep the database unless its a dashboard.
# - Add a "running: pending,yes,done,dead" and show an appropriate icon beside jobs
# - Decide if holding before the main loop until 'systemctl is-system-running' returns 'running' is a good
# idea or not.
# - Write the status of this and the scancore daemon to /etc/anvil/anvil.motd and symlink it to /etc/motd.d/
# - Write a script that runs in crontab at UTC 17:00 that sends an email if Scancore or anvil-daemon are disabled.
# - Examine limites in: https://www.freedesktop.org/software/systemd/man/systemd.exec.html#LimitCPU=
# - Write a background program to scan the BCN and uses OUI data to try and find / auto-configure PDUs and UPSes
# -
# - Increase DRBD's default timeout
# - Check for and enable persistent journald logging
#
# NOTE:
# - For later; 'reboot --force --force' immediately kills the OS, like disabling ACPI on EL6 and hitting the
# power button. Might be useful in ScanCore down the road.
#
# Switches:
#
# --main-loop-only
#
# This skips the one-time, start-up tasks and just goes into the main-loop,
#
# --no-start
#
# This will prevent any pending jobs from being picked up and started in this run. Note that other job checks will still happen.
#
# --refresh-json
#
# This just updates the JSON files used by the web interface. It is the same as '--run-once --main-loop-only --no-start'
#
# --run-once
#
# This will tell the program to exit after runn the main loop once.
#
# --startup-only
#
# This will tell the program to exit after running the start up tasks, so the main loop won't run.
#
use strict;
use warnings;
use Anvil::Tools;
use Proc::Simple;
#use Time::HiRes qw ( time sleep );
use JSON;
use HTML::Strip;
use HTML::FromText;
use Data::Dumper;
use Text::Diff;
my $THIS_FILE = ($0 =~ /^.*\/(.*)$/)[0];
my $running_directory = ($0 =~ /^(.*?)\/$THIS_FILE$/)[0];
if (($running_directory =~ /^\./) && ($ENV{PWD}))
{
$running_directory =~ s/^\./$ENV{PWD}/;
}
# Turn off buffering so that the pinwheel will display while waiting for the SSH call(s) to complete.
$| = 1;
# Prevent a discrepency between UID/GID and EUID/EGID from throwing an error.
$< = $>;
$( = $);
# NOTE: Setting 'log_level' and 'log_secure' here will get overridden in the main lopp. Use the Log methods
# in the loop as well to override defaults in code.
my $anvil = Anvil::Tools->new();
# Make sure we're running as 'root'
# $< == real UID, $> == effective UID
if (($< != 0) && ($> != 0))
{
# Not root
print $anvil->Words->string({key => "error_0005"})."\n";
$anvil->nice_exit({exit_code => 1});
}
# If, so some reason, anvil.conf is lost, create it.
$anvil->System->_check_anvil_conf();
# Connect to the database(s). If we have no connections, we'll proceed anyway as one of the 'run_once' tasks
# is to setup the database server.
$anvil->Database->connect({check_if_configured => 1, check_for_resync => 1});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0132"});
# If I have no databases, sleep for a second and then exit (systemd will restart us).
if (not $anvil->data->{sys}{database}{connections})
{
# If this is a dashboard, try to configure and then connect to the local database. If this isn't a
# dashboard, then just go into a loop waiting for a database to be configured.
if ($anvil->Get->host_type eq "striker")
{
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0201"});
prep_database($anvil);
# Try connecting again
$anvil->refresh();
$anvil->Database->connect();
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0132"});
if (not $anvil->data->{sys}{database}{connections})
{
# Still nothing, sleep and exit.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, 'print' => 1, level => 0, key => "error_0003"});
$anvil->nice_exit({exit_code => 2});
}
}
else
{
# Wait until we have one.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, 'print' => 1, level => 0, key => "error_0075"});
until($anvil->data->{sys}{database}{connections})
{
sleep 10;
check_network($anvil);
$anvil->refresh();
$anvil->Database->connect({check_if_configured => 1, check_for_resync => 1});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 3, key => "log_0132"});
if (not $anvil->data->{sys}{database}{connections})
{
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, 'print' => 0, level => 3, key => "log_0439"});
}
}
}
}
# Read switches
$anvil->data->{switches}{'refresh-json'} = "";
$anvil->data->{switches}{'run-once'} = 0;
$anvil->data->{switches}{'main-loop-only'} = 0;
$anvil->data->{switches}{'no-start'} = 0;
$anvil->data->{switches}{'startup-only'} = 0;
$anvil->Get->switches;
$anvil->Log->level({set => 2});
$anvil->Log->secure({set => 1});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 2, key => "log_0115", variables => { program => $THIS_FILE }});
if ($anvil->data->{switches}{'refresh-json'})
{
$anvil->data->{switches}{'run-once'} = 1;
$anvil->data->{switches}{'main-loop-only'} = 1;
$anvil->data->{switches}{'no-start'} = 1;
}
# This is used to track initial checkes / repairs of network issues.
$anvil->data->{sys}{network}{initial_checks} = 0;
# There are some things we only want to run on (re)start and don't need to always run.
run_once($anvil) if not $anvil->data->{switches}{'main-loop-only'};
# Calculate my sum so that we can exit if it changes later.
$anvil->Storage->record_md5sums;
# What time is it, Mr. Fox?
my $now_time = time;
# To avoid multiple dashboards running a network scan and OUI parse, the dashboard peer with the lowest
# host_uuid sets it's daily checks to run now, and the other(s) will get a two hour's delay.
my $delay = set_delay($anvil);
# Once a minute, we'll check the md5sums and see if we should restart.
# Once a day, we'll refresh an Install Target's RPM repository (has no effect on non-Striker dashboards).
$anvil->data->{timing}{minute_checks} = 60;
$anvil->data->{timing}{daily_checks} = 86400;
$anvil->data->{timing}{repo_update_interval} = 86400;
$anvil->data->{timing}{next_minute_check} = $now_time - 1;
$anvil->data->{timing}{next_daily_check} = ($now_time + $delay) - 1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
"s1:timing::minute_checks" => $anvil->data->{timing}{minute_checks},
"s2:timing::daily_checks" => $anvil->data->{timing}{daily_checks},
"s3:timing::repo_update_interval" => $anvil->data->{timing}{repo_update_interval},
"s4:now_time" => $now_time,
"s5:delay" => $delay,
"s6:timing::next_minute_check" => $anvil->data->{timing}{next_minute_check},
"s7:timing::next_daily_check" => $anvil->data->{timing}{next_daily_check},
}});
# Disconnect. We'll reconnect inside the loop
$anvil->Database->disconnect();
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 3, key => "log_0203"});
# This will prevent restarting while jobs are running.
$anvil->data->{sys}{jobs_running} = 0;
# When we periodically check if system files have changed, we'll also ask Database>connect() to check if it
# needs to be configured or updated. This is done periodically as it is expensive to run on every loop.
my $check_if_database_is_configured = 0;
# These are the things we always want running.
while(1)
{
# Reload defaults, re-read the config and then connect to the database(s)
$anvil->refresh();
# If, so some reason, anvil.conf is lost, create it.
$anvil->System->_check_anvil_conf();
$anvil->Database->connect({check_if_configured => $check_if_database_is_configured, check_for_resync => 1});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 3, key => "log_0132"});
# Mark that we don't want to check the database now.
$check_if_database_is_configured = 0;
# If this host is mapping the network, we'll skip a lot of stuff. If set for over an hour, we'll
# clear it.
$anvil->data->{sys}{mapping_network} = check_if_mapping($anvil);
if ($anvil->data->{sys}{database}{connections})
{
# Run the normal tasks
keep_running($anvil);
# Handle periodic tasks
handle_periodic_tasks($anvil) if not $anvil->data->{sys}{mapping_network};
}
else
{
# No databases available, we'll update the state file in case this host is having it's
# network mapped and the interface used to talk to the databases went down. That's all we
# can do though.
update_state_file($anvil);
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 0, key => "log_0202"});
}
# Exit if 'run-once' selected.
if ($anvil->data->{switches}{'run-once'})
{
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 0, priority => "alert", key => "message_0055"});
$anvil->nice_exit({exit_code => 0});
}
# Disconnect from the database(s) and sleep now.
$anvil->Database->disconnect();
sleep(2);
}
$anvil->nice_exit({exit_code => 0});
#############################################################################################################
# Functions #
#############################################################################################################
# Check to see if we're mapping the network on this host.
sub check_if_mapping
{
my ($anvil) = @_;
$anvil->data->{sys}{mapping_network} = 0;
if ($anvil->data->{sys}{database}{connections})
{
my ($map_network_value, $map_network_uuid, $map_network_mtime, $map_network_modified_date) = $anvil->Database->read_variable({
debug => 3,
variable_name => "config::map_network",
variable_source_table => "hosts",
variable_source_uuid => $anvil->data->{sys}{host_uuid},
});
# We'll run for a day (should be cancelled by the program when the user's done, so this
# shouldn't fire in practice).
my $expire_age = 86400;
my $map_network_age = 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
's1:map_network_value' => $map_network_value,
's2:map_network_mtime' => $map_network_mtime,
's3:map_network_modified_date' => $map_network_modified_date,
's4:map_network_uuid' => $map_network_uuid,
}});
if ($map_network_uuid)
{
$map_network_age = time - $map_network_mtime;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { map_network_age => $map_network_age }});
}
if ($map_network_value)
{
# How long ago was it set?
$anvil->data->{switches}{'clear-mapping'} = "" if not defined $anvil->data->{switches}{'clear-mapping'};
if (($map_network_age >= $expire_age) or ($anvil->data->{switches}{'clear-mapping'}))
{
# Clear it.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0470"});
$anvil->Database->insert_or_update_variables({
debug => 3,
variable_value => 0,
variable_uuid => $map_network_uuid,
update_value_only => 1,
});
}
else
{
# Mark it so we only track the network.
my $say_age = $anvil->Convert->add_commas({number => $expire_age});
my $timeout = $anvil->Convert->add_commas({number => ($expire_age - $map_network_age)});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0471", variables => {
age => $say_age,
timeout => $timeout,
}});
$anvil->data->{sys}{mapping_network} = 1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { "sys::mapping_network" => $anvil->data->{sys}{mapping_network} }});
# Close any open ssh connections.
foreach my $ssh_fh_key (keys %{$anvil->data->{cache}{ssh_fh}})
{
my $ssh_fh = $anvil->data->{cache}{ssh_fh}{$ssh_fh_key};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
ssh_fh_key => $ssh_fh_key,
ssh_fh => $ssh_fh,
}});
if ($ssh_fh =~ /^Net::OpenSSH/)
{
$ssh_fh->disconnect();
}
delete $anvil->data->{cache}{ssh_fh}{$ssh_fh_key};
}
}
}
}
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { "sys::mapping_network" => $anvil->data->{sys}{mapping_network} }});
return($anvil->data->{sys}{mapping_network});
}
# This decides if the local system will delay daily runs on start-up.
sub set_delay
{
my ($anvil) = @_;
my $delay = 7200;
my $type = $anvil->Get->host_type();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { type => $type }});
if ($type eq "striker")
{
foreach my $uuid (keys %{$anvil->data->{database}})
{
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
"sys::host_uuid" => $anvil->data->{sys}{host_uuid},
uuid => $uuid,
}});
if ($uuid eq $anvil->data->{sys}{host_uuid})
{
$delay = 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { delay => $delay }});
}
last;
}
}
else
{
# Not a dashboard, don't delay
$delay = 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { delay => $delay }});
}
return($delay);
}
# This checks to see if it's time to see if the network is ok and, if the system has been up long enough,
# checks and tries to repair network issues.
sub check_network
{
my ($anvil) = @_;
# The network sometimes doesn't come up, but we don't want to try recovering it too soon. As such,
# we'll start watching the network after the uptime is 2 minutes.
my $uptime = $anvil->Get->uptime;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { uptime => $uptime }});
if ($uptime > 120)
{
# Check that bonds are up. Degraded bonds will be left alone.
if (not $anvil->data->{sys}{network}{initial_checks})
{
my $running = $anvil->System->check_daemon({daemon => "NetworkManager"});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { running => $running }});
if (not $running)
{
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 2, key => "message_0250", variables => { daemon => "NetworkManager" }});
my $return_code = $anvil->System->start_daemon({daemon => "NetworkManager"});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { return_code => $return_code }});
}
#$anvil->Network->check_network({heal => "all"});
$anvil->data->{sys}{network}{initial_checks} = 1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
"sys::network::initial_checks" => $anvil->data->{sys}{network}{initial_checks},
}});
}
else
{
### NOTE: This is constantly trying to "fix" healthy bonds, without a know way to
### trigger to debug. As such, disabling for now.
#$anvil->Network->check_network({heal => "down_only"});
}
check_firewall($anvil);
}
return(0);
}
# This handles running tasks that only run on some loops.
sub handle_periodic_tasks
{
my ($anvil) = @_;
my $now_time = time;
my $type = $anvil->Get->host_type();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
"s1:now_time" => $now_time,
"s2:timing::next_minute_check" => $anvil->data->{timing}{next_minute_check},
"s3:timing::next_daily_check" => $anvil->data->{timing}{next_daily_check},
"s4:type" => $type,
}});
# Time to run once per minute tasks.
if ($now_time >= $anvil->data->{timing}{next_minute_check})
{
# Check the firewall needs to be updated.
check_network($anvil);
# Check to see if the PXE environment needs to be updated.
check_install_target($anvil);
# Check that the users we care about have ssh public keys and they're recorded in ssh_keys.
$anvil->System->check_ssh_keys({debug => 3});
$anvil->System->update_hosts({debug => 3});
# Check if the files on disk have changed. Even if it is time to check, don't if a job is
# running.
if ((not $anvil->data->{timing}{jobs_running}) && ($anvil->Storage->check_md5sums))
{
# NOTE: We exit with '0' to prevent systemctl from showing a scary red message.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 0, priority => "alert", key => "message_0014"});
$anvil->nice_exit({exit_code => 0});
}
# Mark that we want to check the database config next time.
$check_if_database_is_configured = 1;
# Update the next check time.
$anvil->data->{timing}{next_minute_check} = $now_time + $anvil->data->{timing}{minute_checks};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
"s1:timing::minute_checks" => $anvil->data->{timing}{minute_checks},
"s2:timing::next_minute_check" => $anvil->data->{timing}{next_minute_check},
}});
# Even when this runs, it should finish in under ten seconds so we don't need to background it.
my ($parse_output, $return_code) = $anvil->System->call({debug => 3, shell_call => $anvil->data->{path}{exe}{'anvil-parse-fence-agents'}.$anvil->Log->switches, source => $THIS_FILE, line => __LINE__});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { parse_output => $parse_output }});
# Scan the local network.
update_state_file($anvil);
# Make sure the shared directories exist.
foreach my $target (sort {$a cmp $b} keys %{$anvil->data->{path}{directories}{shared}})
{
my $directory = $anvil->data->{path}{directories}{shared}{$target};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
target => $target,
directory => $directory,
}});
if (not -e $anvil->data->{path}{directories}{shared}{$target})
{
my $failed = $anvil->Storage->make_directory({
directory => $directory,
group => "apache",
user => "apache",
mode => "0775",
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { failed => $failed }});
if ($failed)
{
# Something went wrong.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 0, priority => "alert", key => "log_0254", variables => {
directory => $directory,
}});
}
else
{
# Success
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0255", variables => {
directory => $directory,
}});
}
}
}
# Check mail server config.
my $problem = $anvil->Email->check_config({debug => 3});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { problem => $problem }});
}
# Now check to see if it's time to run daily tasks.
if ($now_time >= $anvil->data->{timing}{next_daily_check})
{
### NOTE: We call it once/day, but this will also trigger on restart of anvil-daemon. As such, we
### don't use '--force' and let striker-manage-install-target skip the repo update if it happened
### recently enough.
if ($type eq "striker")
{
# Age out old data. This takes up to a minute.
$anvil->Database->_age_out_data();
# Archive old data
$anvil->Database->archive_database();
# Record a job, don't call it directly. It takes too long to run.
my ($job_uuid) = $anvil->Database->insert_or_update_jobs({
file => $THIS_FILE,
line => __LINE__,
job_command => $anvil->data->{path}{exe}{'striker-manage-install-target'}." --refresh".$anvil->Log->switches,
job_data => "",
job_name => "install-target::refresh",
job_title => "job_0015",
job_description => "job_0017",
job_progress => 0,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { job_uuid => $job_uuid }});
# Update the OUI data.
($job_uuid) = $anvil->Database->insert_or_update_jobs({
file => $THIS_FILE,
line => __LINE__,
job_command => $anvil->data->{path}{exe}{'striker-parse-oui'}.$anvil->Log->switches,
job_data => "",
job_name => "oui-data::refresh",
job_title => "job_0064",
job_description => "job_0065",
job_progress => 0,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { job_uuid => $job_uuid }});
# Scan the networks
($job_uuid) = $anvil->Database->insert_or_update_jobs({
file => $THIS_FILE,
line => __LINE__,
job_command => $anvil->data->{path}{exe}{'striker-scan-network'}.$anvil->Log->switches,
job_data => "",
job_name => "scan-network::refresh",
job_title => "job_0066",
job_description => "job_0067",
job_progress => 0,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { job_uuid => $job_uuid }});
}
# Update the next check time.
$anvil->data->{timing}{next_daily_check} = $now_time + $anvil->data->{timing}{daily_checks};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
"s1:timing::daily_checks" => $anvil->data->{timing}{daily_checks},
"s2:timing::next_daily_check" => $anvil->data->{timing}{next_daily_check},
}});
}
return(0);
}
# This calls striker-manage-install-target to see if the dhcpd is running or not. If it is or isn't, the config
# variable 'install-target::enabled' is set/updated. On non-Striker hosts, this simply returns without doing
# anything.
sub check_install_target
{
my ($anvil) = @_;
my $system_type = $anvil->Get->host_type();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { system_type => $system_type }});
if ($system_type ne "striker")
{
# Not a dashboard, nothing to do.
return(0);
}
my $status = "unavailable";
my ($output, $return_code) = $anvil->System->call({shell_call => $anvil->data->{path}{exe}{'striker-manage-install-target'}." --status --check --no-refresh".$anvil->Log->switches});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { output => $output }});
foreach my $line (split/\n/, $output)
{
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { line => $line }});
if ($line =~ /status=(\d)/)
{
my $digit = $1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { digit => $digit }});
if ($digit == 0)
{
$status = "disabled";
}
elsif ($digit == 1)
{
$status = "enabled";
}
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { status => $status }});
last;
}
}
# Record the status
$anvil->Database->insert_or_update_variables({
variable_name => "install-target::enabled",
variable_source_uuid => $anvil->Get->host_uuid,
variable_source_table => "hosts",
variable_value => $status,
variable_default => "unavailable",
variable_description => "striker_0110",
variable_section => "system",
});
return(0);
}
# These are tools that don't need to constantly run. They'll typically run when the server starts up or the
# daemon is restarted or reloaded.
sub run_once
{
my ($anvil) = @_;
# Check that the database is ready.
prep_database($anvil);
# Check to see if we need to do boot-time tasks. We only run these if we've just booted
boot_time_tasks($anvil);
# Check the ssh stuff.
# NOTE: This actually runs again in the minutes tasks, but needs to run on boot as well.
$anvil->System->check_ssh_keys();
# Check setuid wrappers
check_setuid_wrappers($anvil);
# Check journald is configured for persistent storage.
check_journald($anvil);
if ($anvil->data->{switches}{'startup-only'})
{
$anvil->nice_exit({exit_code => 0});
}
return(0);
}
sub check_journald
{
my ($anvil) = @_;
# Check the journald.conf to ensure logging in configured to be persistent.
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { 'path::configs::journald.conf' => $anvil->data->{path}{configs}{'journald.conf'} }});
my $peristent_seen = 0;
my $change_storage = 0;
my $old_journald_conf = $anvil->Storage->read_file({file => $anvil->data->{path}{configs}{'journald.conf'}});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { old_journald_conf => $old_journald_conf }});
foreach my $line (split/\n/, $old_journald_conf)
{
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { line => $line }});
if ($line =~ /^Storage=(.*)$/)
{
my $value = $1;
if ($value eq "persistent")
{
$peristent_seen = 1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { peristent_seen => $peristent_seen }});
}
else
{
$change_storage = 1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { change_storage => $change_storage }});
}
}
}
# Make sure the journald directory
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { 'path::directories::journald' => $anvil->data->{path}{directories}{journald} }});
if (not -d $anvil->data->{path}{directories}{journald})
{
$anvil->Storage->make_directory({
debug => 2,
directory => $anvil->data->{path}{directories}{journald},
});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "message_0248", variables => { directory => $anvil->data->{path}{directories}{journald} }});
}
# Make sure the journald is configured for persistent storage.
if (not $peristent_seen)
{
my $storage_added = 0;
my $new_journald_conf = "";
foreach my $line (split/\n/, $old_journald_conf)
{
if (($line =~ /^Storage=/) && ($change_storage))
{
if (not $storage_added)
{
$storage_added = 1;
$new_journald_conf .= "Storage=persistent\n";
}
next;
}
if (($line =~ /^#Storage=/) && (not $storage_added))
{
$storage_added = 1;
$new_journald_conf .= "Storage=persistent\n";
}
$new_journald_conf .= $line."\n";
}
if (not $storage_added)
{
$new_journald_conf .= "Storage=persistent\n";
}
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { new_journald_conf => $new_journald_conf }});
$anvil->Storage->write_file({
debug => 3,
secure => 0,
file => $anvil->data->{path}{configs}{'journald.conf'},
body => $new_journald_conf,
mode => "0644",
overwrite => 1,
});
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "message_0013", variables => { file => $anvil->data->{path}{configs}{'journald.conf'} }});
# Restart the journald service.
my $shell_call = $anvil->data->{path}{exe}{systemctl}." restart systemd-journald.service";
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { shell_call => $shell_call }});
my ($output, $return_code) = $anvil->System->call({debug => 3, shell_call => $shell_call, source => $THIS_FILE, line => __LINE__});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
output => $output,
return_code => $return_code,
}});
}
return(0);
}
# This creates, as needed, the setuid wrappers used by apache to make certain system calls.
sub check_setuid_wrappers
{
my ($anvil) = @_;
my $host_type = $anvil->Get->host_type();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { host_type => $host_type }});
if ($host_type ne "striker")
{
# Not a dashboard, setuid scripts aren't needed.
return(0);
}
# Does the call_striker-get-peer-data wrapper exist yet?
if (-e $anvil->data->{path}{exe}{'call_striker-get-peer-data'})
{
# Exists, skipping.
* Got the node/dr host initialization form to the point where it can test access and decide if it should show the Red Hat account form. Decided that for M3, node/dr host setup will now be a four-stage process; initial install (over PXE), initialization (install the proper anvil-{node,dr} RPM and connect to the database), setup/map the network, and then add to an Anvil! pair. * Updated striker to no longer try to SSH to a remote machine. To enable this, we'd have to give apache a shell and an SSH key, which is dumb and dangerous when considered. * Created tools/striker-get-peer-data which is meant to be invoked as the 'admin' user (via a setuid c-wrapper). It collects basic data about a target machine and reports what it finds on STDOUT. It gets the password for the target via the database. * Updated anvil-daemon to check/create/update setuid c-wrapper(s), which for now is limited to call_striker-initialize-host. * Created Anvil/Tools/Striker.pm to store Striker web-specific methods, including get_peer_data() which calls tools/striker-initialize-host via the setuid admin call_striker-initialize-host c-wrapper. * In order to allow striker via apache to read a peer's anvil.version, which it can no longer do over SSH, any connection to a peer where the anvil.version is read is cached as /etc/anvil/anvil.<peer>.version. When Get->anvil_version is called as 'apache', this file is read instead. * Updated Database->resync_databases() and ->_find_behind_databases() to ignore the 'states' table. * Created tools/striker-initialize-host which will be called as a job to initialize a node/dr host. Signed-off-by: Digimer <digimer@alteeve.ca>
5 years ago
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 3, key => "log_0436", variables => { wrapper => $anvil->data->{path}{exe}{'call_striker-get-peer-data'} }});
}
else
{
# What is the admin user and group ID?
my $admin_uid = getpwnam('admin');
my $admin_gid = getgrnam('admin');
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
admin_uid => $admin_uid,
admin_gid => $admin_gid,
}});
next if not $admin_uid;
next if not $admin_gid;
# Write the body out
my $call_striker_get_peer_data_body = "#define REAL_PATH \"".$anvil->data->{path}{exe}{'striker-get-peer-data'}."\"\n";
$call_striker_get_peer_data_body .= "main(ac, av)\n";
$call_striker_get_peer_data_body .= "char **av;\n";
$call_striker_get_peer_data_body .= "{\n";
$call_striker_get_peer_data_body .= " setuid(".$admin_uid.");\n";
$call_striker_get_peer_data_body .= " setgid(".$admin_gid.");\n";
$call_striker_get_peer_data_body .= " execv(REAL_PATH, av);\n";
$call_striker_get_peer_data_body .= "}\n";
my $error = $anvil->Storage->write_file({
debug => 3,
file => $anvil->data->{path}{exe}{'call_striker-get-peer-data'}.".c",
body => $call_striker_get_peer_data_body,
mode => '644',
overwrite => 1,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { error => $error }});
# If it wrote out, compile it.
if (not -e $anvil->data->{path}{exe}{'call_striker-get-peer-data'}.".c")
{
# Failed to write.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, priority => "alert", key => "error_0071", variables => { file => $anvil->data->{path}{exe}{'call_striker-get-peer-data'}.".c" }});
}
else
{
# Compile it
my ($output, $return_code) = $anvil->System->call({
debug => 3,
shell_call => $anvil->data->{path}{exe}{gcc}." -o ".$anvil->data->{path}{exe}{'call_striker-get-peer-data'}." ".$anvil->data->{path}{exe}{'call_striker-get-peer-data'}.".c",
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
output => $output,
return_code => $return_code,
}});
# If it compiled, setuid it.
if (not -e $anvil->data->{path}{exe}{'call_striker-get-peer-data'})
{
# Something went wrong compiling it.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, priority => "alert", key => "error_0072", variables => { file => $anvil->data->{path}{exe}{'call_striker-get-peer-data'}.".c" }});
}
else
{
$anvil->Storage->change_owner({
debug => 3,
path => $anvil->data->{path}{exe}{'call_striker-get-peer-data'},
user => 'root',
group => 'root',
});
$anvil->Storage->change_mode({
debug => 3,
path => $anvil->data->{path}{exe}{'call_striker-get-peer-data'},
mode => '4755',
});
}
}
}
return(0);
}
# Configure/update the firewall.
sub check_firewall
{
my ($anvil) = @_;
# Don't call this if we're not configured yet.
my $configured = $anvil->System->check_if_configured({debug => 3});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { configured => $configured }});
# Check the firewall needs to be updated.
if ($configured)
{
my ($output, $return_code) = $anvil->System->call({shell_call => $anvil->data->{path}{exe}{'anvil-manage-firewall'}.$anvil->Log->switches});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { output => $output, return_code => $return_code }});
}
return(0);
}
# This handles tasks that need to run on boot (if any)
sub boot_time_tasks
{
my ($anvil) = @_;
# If the uptime is less than ten minutes, clear the reboot flag.
my $uptime = $anvil->Get->uptime;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { uptime => $uptime }});
# Now find out if a reboot is listed as needed and when it was last changed.
my $reboot_needed = 0;
my $changed_seconds_ago = 0;
my $query = "
SELECT
variable_value,
(SELECT extract(epoch from now()) - extract(epoch from modified_date)) AS changed_seconds_ago
FROM
variables
WHERE
variable_source_table = 'hosts'
AND
variable_source_uuid = ".$anvil->Database->quote($anvil->Get->host_uuid)."
AND
variable_name = 'reboot::needed'
;";
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 2, key => "log_0124", variables => { query => $query }});
my $results = $anvil->Database->query({query => $query, source => $THIS_FILE, line => __LINE__});
my $count = @{$results};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
results => $results,
count => $count,
}});
if ($count)
{
$reboot_needed = $results->[0]->[0];
$changed_seconds_ago = $results->[0]->[1];
$changed_seconds_ago =~ s/^(\d+)\..*$/$1/;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
reboot_needed => $reboot_needed,
changed_seconds_ago => $changed_seconds_ago,
}});
}
# If a reboot is needed, see if the uptime is less than the time since the reboot needed flag was
# set. If the uptime is less, then the system rebooted since it was requested so clear it. h/t to
# Lisa Seelye (@thedoh) for this idea!
my $difference = ($changed_seconds_ago - $uptime);
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
"s1:reboot_needed" => $reboot_needed,
"s2:changed_seconds_ago" => $changed_seconds_ago,
"s3:uptime" => $uptime,
"s4:difference" => $difference,
}});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
if ($reboot_needed)
{
if ($uptime < $changed_seconds_ago)
{
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
# Clear the reboot request.
$reboot_needed = $anvil->System->reboot_needed({debug => 2, set => 0});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { reboot_needed => $reboot_needed }});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
# Check to see if there was a reboot job in progress. If so, finish it off.
my $job_uuid = $anvil->Job->get_job_uuid({
debug => 2,
program => "anvil-manage-power",
incomplete => 1,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { job_uuid => $job_uuid }});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
if ($job_uuid)
{
# Update the percentage to '100' and then clear the old PID.
my $date_time = $anvil->Get->date_and_time();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { date_time => $date_time }});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
$anvil->Job->update_progress({
progress => 100,
message => "message_0064,!!date_and_time!".$date_time."!!",
job_uuid => $job_uuid,
picked_up_by => 0,
});
}
}
}
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
else
{
# Update our status
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, 'print' => 0, level => 2, key => "log_0572"});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
$anvil->Database->get_hosts({debug => 2});
my $host_uuid = $anvil->Get->host_uuid({debug => 2});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { host_uuid => $host_uuid }});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
$anvil->Database->insert_or_update_hosts({
debug => 2,
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
host_ipmi => $anvil->data->{hosts}{host_uuid}{$host_uuid}{host_ipmi},
host_key => $anvil->data->{hosts}{host_uuid}{$host_uuid}{host_key},
host_name => $anvil->data->{hosts}{host_uuid}{$host_uuid}{host_name},
host_type => $anvil->data->{hosts}{host_uuid}{$host_uuid}{host_type},
host_uuid => $host_uuid,
host_status => "online",
});
# Make sure our stop reason is cleared.
my $variable_uuid = $anvil->Database->insert_or_update_variables({
variable_name => 'system::stop_reason',
variable_value => '',
variable_default => '',
variable_description => 'striker_0279',
variable_section => 'system',
The core logic is done!!!! Still need to finish end-points for the WebUI to hook into, but the core of M3 is complete! Many, many bugs are expected, of course. :) * Created DRBD->check_if_syncsource() and ->check_if_synctarget() that return '1' if the target host is currently SyncSource or SyncTarget for any resource, respectively. * Updated DRBD->update_global_common() to return the unified-format diff if any changes were made to global-common.conf. * Created ScanCore->check_health() that returns the health score for a host. Created ->count_servers() that returns the number of servers on a host, how much RAM is used by those servers and, if available, the estimated migration time of the servers. Updated ->check_temperature() to set/clear/return the time that a host has been in a warning or critical temperature state. * Finished ScanCore->post_scan_analysis_node()!!! It certainly has bugs, and much testing is needed, but the logic is all in place! Oh what a slog that was... It should be far more intelligent than M2 though, once flushed out and tested. * Created Server->active_migrations() that returns '1' if any servers are in a migration on an Anvil! system. Updated ->migrate_virsh() to record how long a migration took in the "server::migration_duration" variable, which is averaged by ScanCore->count_servers() to estimate migration times. * Updated scan-drbd to check/update the global-common.conf file's config at the end of a scan. * Updated ScanCore itself to not scan when in maintenance mode. Also updated it to call 'anvil-safe-start' when ScanCore starts, so long as it is within ten minutes of the host booting. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
variable_source_uuid => $host_uuid,
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
variable_source_table => 'hosts',
});
The core logic is done!!!! Still need to finish end-points for the WebUI to hook into, but the core of M3 is complete! Many, many bugs are expected, of course. :) * Created DRBD->check_if_syncsource() and ->check_if_synctarget() that return '1' if the target host is currently SyncSource or SyncTarget for any resource, respectively. * Updated DRBD->update_global_common() to return the unified-format diff if any changes were made to global-common.conf. * Created ScanCore->check_health() that returns the health score for a host. Created ->count_servers() that returns the number of servers on a host, how much RAM is used by those servers and, if available, the estimated migration time of the servers. Updated ->check_temperature() to set/clear/return the time that a host has been in a warning or critical temperature state. * Finished ScanCore->post_scan_analysis_node()!!! It certainly has bugs, and much testing is needed, but the logic is all in place! Oh what a slog that was... It should be far more intelligent than M2 though, once flushed out and tested. * Created Server->active_migrations() that returns '1' if any servers are in a migration on an Anvil! system. Updated ->migrate_virsh() to record how long a migration took in the "server::migration_duration" variable, which is averaged by ScanCore->count_servers() to estimate migration times. * Updated scan-drbd to check/update the global-common.conf file's config at the end of a scan. * Updated ScanCore itself to not scan when in maintenance mode. Also updated it to call 'anvil-safe-start' when ScanCore starts, so long as it is within ten minutes of the host booting. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { variable_uuid => $variable_uuid }});
* THe get_cpu endpoint was completed. * The get_mmeory endpoint was completed. * The get_replicated_storage endpoint was completed, though it requires testing and likely has issues. To prepare for the get_status endpoint work, I needed to update ScanCore and modules to track the host_status. This commit contains the work needed for this. * Updated ScanCore->post_scan_analysis_striker() to use configured fence devices (except PDUs) to check if a target host is off or on, in there is no host_ipmi interface. In all cases, if a machine can be confirmed on or off, the host_status is now updated. * To support the above fence based power checks, updated scan-cluster to store the on-disk CIB in the new scan_cluster -> scan_cluster_cib colume. * Updated ScanCore->parse_cib() to map stonith primitive IDs to fence agents. Updated ->parse_crm_mon() to not call if the executable doesn't exist to avoid unhelpful error messages in the logs when called from a Striker. * Update DRBD->gather_data() to get the size data from /sys/block/drbd<minor>/size' x '/sys/block/drbd<minor>/queue/logical_block_size so it works when a device is Secondary (and can't be promoted). * Updated Database->get_hosts_info() to record the short host name as well as the stored host name. Created ->update_host_status() as a wrapper to ->insert_or_update_hosts() that only updates the host status. * Updated anvil-join-anvil to disabled ksm and ksmtuned daemons. * Updated scancore and anvil-daemon to set the host_status to 'online' on startup. Signed-off-by: Digimer <digimer@alteeve.ca>
4 years ago
}
# Make sure /etc/hosts is updated.
$anvil->System->update_hosts();
# This handles weird bits for things like bug work-arounds.
handle_special_cases($anvil);
# Now look for jobs that have a job status of 'anvil_startup'
run_jobs($anvil, 1);
# Check the firewall needs to be updated.
check_firewall($anvil);
# If we're a striker, check apache
my $host_type = $anvil->Get->host_type;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { host_type => $host_type }});
if ($host_type eq "striker")
{
$anvil->Striker->check_httpd_conf({debug => 2});
}
return(0);
}
# This handles weird bits for things like bug work-arounds.
sub handle_special_cases
{
my ($anvil) = @_;
# RHBZ #1961562 - https://bugzilla.redhat.com/show_bug.cgi?id=1961562#c16
my $host_type = $anvil->Get->host_type();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { host_type => $host_type }});
if ($host_type ne "striker")
{
# We're a node or DR host. We need to touch this file.
my $work_around_file = "/etc/qemu/firmware/50-edk2-ovmf-cc.json";
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { work_around_file => $work_around_file }});
if (not -e $work_around_file)
{
$anvil->Storage->write_file({
debug => 2,
file => $work_around_file,
body => "",
overwrite => 0,
backup => 0,
mode => "0644",
user => "root",
group => "root",
});
}
}
return(0);
}
# Configure the local database, if needed.
sub prep_database
{
my ($anvil) = @_;
# Only run this if we're a dashboard.
my $host_type = $anvil->Get->host_type();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { host_type => $host_type }});
if ($host_type eq "striker")
{
### NOTE: This failed once, in case / until it happens again, we'll force log level 2 and secure logging.
#my ($database_output, $return_code) = $anvil->System->call({debug => 2, shell_call => $anvil->data->{path}{exe}{'striker-prep-database'}.$anvil->Log->switches, source => $THIS_FILE, line => __LINE__ });
my ($database_output, $return_code) = $anvil->System->call({debug => 2, shell_call => $anvil->data->{path}{exe}{'striker-prep-database'}." -vv --log-secure", source => $THIS_FILE, line => __LINE__ });
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
database_output => $database_output,
return_code => $return_code,
}});
}
return(0);
}
# These are tools that need to keep running.
sub keep_running
{
my ($anvil) = @_;
# Check for jobs that were running and now exited.
if ((not $anvil->data->{sys}{mapping_network}) && (exists $anvil->data->{processes}))
{
foreach my $job_uuid (%{$anvil->data->{jobs}{handles}})
{
# If it's not a handle, delete it.
my $running = $anvil->data->{jobs}{handles}{$job_uuid}->poll();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
"jobs::handles::${job_uuid}" => $anvil->data->{jobs}{handles}{$job_uuid},
running => $running,
}});
# If it's not running, update the table to clear the 'job_picked_up_by' column.
if (not $running)
{
my $exit_status = $anvil->data->{jobs}{handles}{$job_uuid}->exit_status();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 1, list => {
job_uuid => $job_uuid,
exit_status => $exit_status,
}});
# Free up memory
$anvil->data->{jobs}{handles}{$job_uuid}->cleanup();
$anvil->Job->clear({job_uuid => $job_uuid});
}
}
}
# If we're configured, write out the status JSON file. If we're not configured, Update hardware state files.
my $configured = $anvil->System->check_if_configured;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { configured => $configured }});
if ((not $anvil->data->{sys}{mapping_network}) && ($configured))
{
# Write out state information for all known Anvil! systems and the information from
# unconfigured nods and DR hosts, using just database data (hence, fast enough to run
# constantly).
$anvil->System->generate_state_json({debug => 3});
}
else
{
# Run this to monitor the network in real time.
update_state_file($anvil);
}
# Run any pending jobs by calling 'anvil-jobs' with the 'job_uuid' as a background process.
run_jobs($anvil, 0) if not $anvil->data->{sys}{mapping_network};
return(0);
}
# This will check for any jobs that aren't at 100%. For each found, if 'picked_up_by' is set, a check is made
# to see if the PID is still alive. If it isn't, or if 'picked_up_by' is not set, the appropriate tool is
# invoked to handle it.
sub run_jobs
{
my ($anvil, $startup) = @_;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { startup => $startup }});
# This will be set to 1 if any jobs are not complete, preventing a restart of the daemon if it's
# changed on disk.
$anvil->data->{sys}{jobs_running} = 0;
# We'll also update the jobs.json file.
my $jobs_file = "{\"jobs\":[\n";
# Get a list of pending or incomplete jobs.
my $ended_within = $startup ? 1 : 300;
my $return = $anvil->Database->get_jobs({ended_within => $ended_within});
my $count = @{$return};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
ended_within => $ended_within,
'return' => $return,
count => $count,
}});
foreach my $hash_ref (@{$return})
{
my $job_uuid = $hash_ref->{job_uuid};
my $job_command = $hash_ref->{job_command};
my $job_data = $hash_ref->{job_data};
my $job_picked_up_by = $hash_ref->{job_picked_up_by};
my $job_picked_up_at = $hash_ref->{job_picked_up_at};
my $job_updated = $hash_ref->{job_updated};
my $job_name = $hash_ref->{job_name};
my $job_progress = $hash_ref->{job_progress};
my $job_title = $hash_ref->{job_title};
my $job_description = $hash_ref->{job_description};
my $job_status = $hash_ref->{job_status};
my $started_seconds_ago = $job_picked_up_at ? (time - $job_picked_up_at) : 0;
my $updated_seconds_ago = $job_updated ? (time - $job_updated) : 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
job_uuid => $job_uuid,
job_command => $job_command,
job_data => $job_data,
job_picked_up_by => $job_picked_up_by,
job_picked_up_at => $job_picked_up_at,
job_updated => $job_updated,
job_name => $job_name,
job_progress => $job_progress,
job_title => $job_title,
job_description => $job_description,
job_status => $job_status,
started_seconds_ago => $started_seconds_ago,
updated_seconds_ago => $updated_seconds_ago,
}});
# If this is a start-up call, only start jobs whose status is 'anvil_startup'.
if (($startup) && ($job_status ne "anvil_startup"))
{
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 2, key => "log_0639", variables => {
job_uuid => $job_uuid,
job_command => $job_command,
}});
next;
}
if ($job_progress ne "100")
{
$anvil->data->{sys}{jobs_running} = 1;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { "sys::jobs_running" => $anvil->data->{sys}{jobs_running} }});
}
# See if the job was picked up by a now-dead instance.
if ($job_picked_up_by)
{
# Check if the PID is still active.
$anvil->System->pids({ignore_me => 1});
### TODO: Add a check to verify the job isn't hung.
# Skip if this job is in progress.
if (not exists $anvil->data->{pids}{$job_picked_up_by})
{
# If the job is done, just clear the 'job_picked_up_by' and be done.
if ($job_progress ne "100")
{
# It's possible that the job updated to 100% and exited after we
# gathered the job data, so we won't restart until we've seen it not
# running and not at 100% after 5 loops.
if ((not exists $anvil->data->{lost_job_count}{$job_uuid}) or (not defined $anvil->data->{lost_job_count}{$job_uuid}))
{
$anvil->data->{lost_job_count}{$job_uuid} = 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { "lost_job_count::${job_uuid}" => $anvil->data->{lost_job_count}{$job_uuid} }});
}
if ($anvil->data->{lost_job_count}{$job_uuid} > 5)
{
# The previous job is gone, but the job isn't finished. Start it again.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, priority => "alert", key => "warning_0007", variables => {
command => $job_command,
pid => $job_picked_up_by,
percent => $job_progress,
}});
# Clear some variables.
$job_progress = 0;
$job_status = "message_0056";
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
job_progress => $job_progress,
job_status => $job_status,
}});
# Clear the job.
$anvil->Job->clear({debug => 2, job_uuid => $job_uuid});
$anvil->data->{lost_job_count}{$job_uuid} = 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { "lost_job_count::${job_uuid}" => $anvil->data->{lost_job_count}{$job_uuid} }});
}
else
{
$anvil->data->{lost_job_count}{$job_uuid}++;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { "lost_job_count::${job_uuid}" => $anvil->data->{lost_job_count}{$job_uuid} }});
}
}
# Clear the PID
$job_picked_up_by = 0;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { job_picked_up_by => $job_picked_up_by }});
}
}
# Convert the double-banged strings into a proper message.
my $say_title = $job_title ? $anvil->Words->parse_banged_string({key_string => $job_title}) : "";
my $say_description = $job_description ? $anvil->Words->parse_banged_string({key_string => $job_description}) : "";
my $say_status = $job_status ? $anvil->Words->parse_banged_string({key_string => $job_status}) : "";
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
job_title => $job_title,
say_description => $say_description,
say_status => $say_status,
}});
# Make the status HTML friendly. Strip any embedded HTML then encode the text string.
if ($say_status)
{
my $html_strip = HTML::Strip->new();
$say_status = $html_strip->parse($say_status);
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { say_status => $say_status }});
# Now make the resulting text string HTML friendly
my $text_to_html = HTML::FromText->new({
urls => 1,
email => 1,
lines => 1,
});
$say_status = $text_to_html->parse($say_status);
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { say_status => $say_status }});
}
# Add this to the jobs.json file
my $json_string = to_json ({
job_uuid => $job_uuid,
job_command => $job_command,
job_data => $job_data,
job_picked_up_at => $job_picked_up_at,
job_updated => $job_updated,
job_name => $job_name,
job_progress => $job_progress,
job_title => $say_title,
job_description => $say_description,
job_status => $say_status,
started_seconds_ago => $started_seconds_ago,
updated_seconds_ago => $updated_seconds_ago,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { json_string => $json_string }});
$jobs_file .= $json_string.",\n";
# If the job is done, move on.
next if $job_progress eq "100";
next if $anvil->data->{switches}{'no-start'};
# If 'startup' is set, we only care if 'job_status' is 'anvil_startup'
if ((not $startup) && ($say_status eq "anvil_startup"))
{
# Skip this, it will run next time anvil-daemon restarts.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 2, key => "log_0593", variables => {
command => $job_command,
job_uuid => $job_uuid,
}});
next;
}
# If the job is not running, start it.
if (not $job_picked_up_by)
{
my $command = $job_command." --job-uuid ".$job_uuid;
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0210", variables => { command => $command }});
# Have we started this job recently?
if (exists $anvil->data->{jobs}{$job_uuid}{started})
{
my $last_start = time - $anvil->data->{jobs}{$job_uuid}{started};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { last_start => $last_start }});
if ($last_start < 60)
{
# Skip, Started too recently.
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 1, key => "log_0578", variables => {
command => $command,
last_start => $last_start,
}});
next;
}
}
# Start the job, appending '--job-uuid' to the command.
($anvil->data->{jobs}{handles}{$job_uuid}, my $return_code) = $anvil->System->call({
background => 1,
stdout_file => "/tmp/anvil.job.".$job_uuid.".stdout",
stderr_file => "/tmp/anvil.job.".$job_uuid.".stderr",
shell_call => $command,
source => $THIS_FILE,
line => __LINE__,
});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => {
"jobs::handles::${job_uuid}" => $anvil->data->{jobs}{handles}{$job_uuid},
return_code => $return_code,
}});
# Log the PID (the job should update the database).
my $pid = $anvil->data->{jobs}{handles}{$job_uuid}->pid();
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { pid => $pid }});
# Record that we've tried to start this job, so that we don't try to restart it for any reason for at least a minute.
$anvil->data->{jobs}{$job_uuid}{started} = time;
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { 'jobs::$job_uuid::started' => $anvil->data->{jobs}{$job_uuid}{started} }});
}
}
# Close the jobs file.
$jobs_file =~ s/,\n$/\n/ms;
$jobs_file .= "]}\n";
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { jobs_file => $jobs_file }});
# Write the JSON file
my $output_json = $anvil->data->{path}{directories}{html}."/status/jobs.json";
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => { output_xml => $output_json }});
$anvil->Storage->write_file({
file => $output_json,
body => $jobs_file,
overwrite => 1,
backup => 0,
mode => "0644",
user => "apache",
group => "apache",
});
return(0);
}
# This calls 'anvil-update-states' which will scan the local machine's state (hardware and software) and
# record write it out to an HTML file
sub update_state_file
{
my ($anvil) = @_;
$anvil->Log->entry({source => $THIS_FILE, line => __LINE__, level => 3, key => "log_0480"});
#my $shell_call = $anvil->data->{path}{exe}{'anvil-update-states'}.$anvil->Log->switches;
my $shell_call = $anvil->data->{path}{exe}{'anvil-update-states'};
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 2, list => { shell_call => $shell_call }});
my ($states_output, $return_code) = $anvil->System->call({debug => 3, shell_call => $shell_call, source => $THIS_FILE, line => __LINE__});
$anvil->Log->variables({source => $THIS_FILE, line => __LINE__, level => 3, list => {
states_output => $states_output,
return_code => $return_code,
}});
return(0);
}