如何利用Vultr的API自动更换故障主机

By minirplus on 2019-01-14 in How To

Vultr的Reserved IP要$3/month，Additional IPv4 IP要$2/month，实在是消费不起，手动更换烦了，遂写了个自动程序，来自动更换主机。

想法是这样的，利用定时任务，每隔5分钟检查一次tcp连接状态，3次异常之后，开始进行切换操作。

切换流程

获取主机列表
停用故障主机
创建故障主机快照
创建新主机
用故障主机快照恢复新主机
删除故障主机

测试通过自动化操作，15分钟内自动确认TCP异常，1小时之内就可以自动切换到新主机。

技术细节

在切换过程中有部分操作需要等待执行完成，比如创建快照和从快照恢复，所以使用了swich来进行步骤计数并写入json文件，初始json计数为0，每操作成功一步，增加一位数。好处是当后一步发现之前的结果出错时可以很灵活的进行回滚。

Vultr的API有一个坑，虽然在创建主机的时候提供了snapshot的选项，但是并不会使用这个snapshot创建主机，导致创建出的主机都是毛坯的，还以为是快照出错了。所以还得分成两步，先创建一个空白主机，再从快照恢复，不过恢复后会自动启动主机，这样就少了一个启动的步骤也算是人性化。

在CLI执行ping命令的时候由于会一直ping下去，程序会进入死循环，在命令前加上timeout就可以限制命令执行时间，不过很可惜，在php中执行命令并没有用。

公用函数

注意替换高亮行的API-Key、缓存路径

function callAPI($method, $url, $data){

$curl = curl_init();

switch ($method){

case "POST":

curl_setopt($curl, CURLOPT_POST, 1);

if ($data)

curl_setopt($curl, CURLOPT_POSTFIELDS, $data);

break;

case "PUT":

curl_setopt($curl, CURLOPT_CUSTOMREQUEST, "PUT");

if ($data)

curl_setopt($curl, CURLOPT_POSTFIELDS, $data);

break;

default:

if ($data)

$url = sprintf("%s?%s", $url, http_build_query($data));

}

// OPTIONS:

curl_setopt($curl, CURLOPT_URL, $url);

curl_setopt($curl, CURLOPT_HTTPHEADER, array(

'API-Key: your_vultr_api_key',

));

curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1);

curl_setopt($curl, CURLOPT_HTTPAUTH, CURLAUTH_BASIC);

// EXECUTE:

$result = curl_exec($curl);

if(!$result){

$httpcode = curl_getinfo($curl, CURLINFO_HTTP_CODE);

curl_close($curl);

return $httpcode;

}

curl_close($curl);

return $result;

}

if ( ! function_exists( 'array_key_last' ) ) {

function array_key_last( $array ) {

$key = NULL;

if ( is_array( $array ) ) {

end( $array );

$key = key( $array );

}

return $key;

}

if (!function_exists('array_key_first')) {

function array_key_first(array $array)

{

if (count($array)) {

reset($array);

return key($array);

}

return null;

}

function output_json($step){

$outputArray['step'] = $step;

$outputJSON = json_encode($outputArray);

$fp = fopen('data_auto_swap_ip_vultr.json', 'w') or die("Unable to open file!");

fwrite($fp, $outputJSON);

fclose($fp);

}

function load_json(){

if(file_exists("data_auto_swap_ip_vultr.json")){

$load_json = file_get_contents("data_auto_swap_ip_vultr.json");

}else{

$load_json = '{"step":0}';

echo 'Can not Load last process status, will reset to: '.$load_json;

echo "\n";

}

return json_decode($load_json, true);

}

步骤零：获取主机列表及程序执行状态

echo 'Get Last Process Status...';

echo "\n";

$last_status = load_json();

echo 'Last process status: '.$last_status['step'];

echo "\n";

$get_data = callAPI('GET', 'https://api.vultr.com/v1/server/list', false);

$response = json_decode($get_data, true);

$server_count = count($response);

echo 'Server Number: '.$server_count;

echo "\n";

echo 'Get Server Info...';

echo "\n";

reset($response);

$first_server_id = array_key_first($response);

$first_server_info = $response[array_key_first($response)];

$last_server_id = array_key_last($response);

$last_server_info = $response[array_key_last($response)];

echo 'First Server ID: '.$first_server_id;

echo "\n";

echo 'First Server Status: '.$first_server_info['power_status'];

echo "\n";

echo 'First Server Charges: '.$first_server_info['pending_charges'];

echo "\n";

echo 'Last Server ID: '.$last_server_id;

echo "\n";

echo 'Last Server Status: '.$last_server_info['power_status'];

echo "\n";

echo 'Last Server Charges: '.$last_server_info['pending_charges'];

echo "\n";

使用状态码执行命令

switch ($last_status['step']) {

// insert code below

}

步骤一：停用故障主机

case 0:

echo 'Stop old server...';

echo "\n";

if($first_server_info['power_status'] == "running"){

echo 'Server(ID='.$first_server_id.') is running, will stop it';

echo "\n";

echo 'Stop Server(ID='.$first_server_id.')...';

echo "\n";

$data = "SUBID=".$first_server_id;

$get_data = callAPI('POST', 'https://api.vultr.com/v1/server/halt', $data);

if($get_data = 200){

echo 'Stop Server(ID='.$first_server_id.') Succeed';

echo "\n";

}

die();

}else{

echo 'Old Server(ID='.$first_server_id.') is stopped, will continue...';

echo "\n";

}

output_json($last_status['step']+1);

break;

步骤二：创建快照

注意高亮行，设置快照有效时间，另时区为UTC+8，创建快照需要等待，此步骤会进行状态判断

case 1:

echo 'Create snapshot...';

echo "\n";

// get snapshot list

$get_data = callAPI('GET', 'https://api.vultr.com/v1/snapshot/list', false);

$response = json_decode($get_data, true);

$first_snapshot_id = array_key_first($response);

$first_snapshot_info = $response[array_key_first($response)];

echo 'snapshot ID: '.$first_snapshot_id;

echo "\n";

$snapshot_status = $first_snapshot_info['status'];

echo 'snapshot Status: '.$snapshot_status;

echo "\n";

echo 'snapshot create date: '.date("Y-m-d H:i:s",strtotime($first_snapshot_info['date_created'])+8*60*60);

echo "\n";

$snapshot_time_age = ((time()-(strtotime($first_snapshot_info['date_created'])+8*60*60))/60);

echo 'snapshot time ago: '.$snapshot_time_age.' min';

echo "\n";

$snapshot_time_age_limit = 60; // snapshot time limit

if($snapshot_status == 'complete' and $snapshot_time_age <= $snapshot_time_age_limit){

echo 'snapshot has complete and newest (less than '.$snapshot_time_age_limit.' min), will continue...';

echo "\n";

output_json($last_status['step']+1);

}

if($snapshot_status !== 'complete' and $snapshot_time_age <= $snapshot_time_age_limit){

echo 'snapshot has not complete, will waiting until it complete';

echo "\n";

die();

}

if($snapshot_status == 'complete' and $snapshot_time_age > $snapshot_time_age_limit){

echo 'snapshot has complete but too old, will create a new snapshot';

echo "\n";

// create snapshot

echo 'Create SNAPSHOT...';

echo "\n";

$data = "SUBID=".$first_server_id;

$get_data = callAPI('POST', 'https://api.vultr.com/v1/snapshot/create', $data);

$response = json_decode($get_data, true);

$snapshot_id = $response['SNAPSHOTID'];

echo 'SNAPSHOT ID: '.$snapshot_id;

echo "\n";

echo 'Snapshot is processing, run again to continue...'.$snapshot_id;

echo "\n";

die();

}

break;

步骤三：创建新服务器

注意高亮行，设置新创建服务器的区域、套餐、操作系统，下值为法兰克福、5美元、Debian 9

case 2:

echo 'Create new server...';

echo "\n";

// create new server

if($server_count <= 1){

echo 'Creating New Server...';

echo "\n";

$data_array = array(

"DCID" => 9,

"VPSPLANID" => 201,

"OSID" => 244

);

$get_data = callAPI('POST', 'https://api.vultr.com/v1/server/create', $data_array);

$response = json_decode($get_data, true);

echo 'New Server has Created(id='.$response['SUBID'].')';

echo "\n";

output_json($last_status['step']+1);

}

break;

步骤四：从快照恢复

快照恢复需要等待，此步骤会进行状态判断

case 3:

echo 'Restore snapshot...';

echo "\n";

// get snapshot list

$get_data = callAPI('GET', 'https://api.vultr.com/v1/snapshot/list', false);

$response = json_decode($get_data, true);

$first_snapshot_id = array_key_first($response);

$first_snapshot_info = $response[array_key_first($response)];

echo 'snapshot ID: '.$first_snapshot_id;

echo "\n";

$snapshot_status = $first_snapshot_info['status'];

echo 'snapshot Status: '.$snapshot_status;

echo "\n";

echo 'snapshot create date: '.date("Y-m-d H:i:s",strtotime($first_snapshot_info['date_created'])+8*60*60);

echo "\n";

$snapshot_time_age = ((time()-(strtotime($first_snapshot_info['date_created'])+8*60*60))/60);

echo 'snapshot time ago: '.$snapshot_time_age.' min';

echo "\n";

if($snapshot_status == 'complete' and $last_server_info['power_status'] == 'running'){

$data_array = array(

"SUBID" => $last_server_id,

"SNAPSHOTID" => $first_snapshot_id

);

$get_data = callAPI('POST', 'https://api.vultr.com/v1/server/restore_snapshot', $data_array);

$response = json_decode($get_data, true);

if($response == NULL){

echo 'New Server has not installed';

echo "\n";

die();

}

var_dump($response);

echo "\n";

output_json($last_status['step']+1);

}else{

echo 'snapshot status: '.$snapshot_status.' server status: '.$last_server_info['power_status'];

echo "\n";

}

break;

步骤五：删除旧服务器

case 4:

echo 'Destroy old server...';

echo "\n";

// Destroy old server

if($last_server_info['power_status'] == 'stopped'){

echo 'server restore not complete, waiting for restore complete...';

echo "\n";

}

elseif($server_count > 1){

echo 'Old server(id='.$first_server_id.') exists, will destroy old one';

echo "\n";

echo 'Destroy old server(id='.$first_server_id.')...';

echo "\n";

$data = "SUBID=$first_server_id";

$get_data = callAPI('POST', 'https://api.vultr.com/v1/server/destroy', $data);

var_dump($get_data);

if($get_data = 200){

echo 'Destroy Server(ID='.$first_server_id.') Succeed';

echo "\n";

}

output_json($last_status['step']+1);

}else{

echo 'only one server, destroy process not running, skip destroy process...';

echo "\n";

output_json($last_status['step']+1);

}

break;

以上更换主机的步骤已全部结束，以下为附加功能

步骤六：更新WireGuard节点信息，Ping节点

此步骤ping阶段会进入假死，需要手动ctrl+c结束，目前使用ini_set可以解决这个问题。

替换高亮部分

case 5:

ini_set('max_execution_time', '5');

echo 'Set Wireguard peer info...';

echo "\n";

echo 'Set Wireguard peer ip to: '.$last_server_info['main_ip'];

echo "\n";

$cmd_set_wireguard = "wg set wg0 peer your_peer_public_key endpoint ".$last_server_info['main_ip'].":your_peer_port";

exec($cmd_set_wireguard,$array);

echo 'Ping Peer...';

echo "\n";

$cmd_ping_peer = "ping your_peer_local_ip";

exec($cmd_ping_peer,$array);

output_json($last_status['step']+1);

break;

循环步骤：检查TCP状态

替换高亮行，TCP检测用的是百度的网站robot检测api，连百度都访问不了，那么肯定是被玩坏了。

default:

$max_tcp_test_time = 3;

$cmd_wget = "wget -q -O - https://ziyuan.baidu.com/robots/doanalysis?site=your_domain_name";

exec($cmd_wget,$array);

$wget_result = json_decode($array[0],true);

$tcp_test = $wget_result['status'];

if($tcp_test == 9){

echo "tcp test result is $tcp_test, block by chinafirewall, will test again";

echo "\n";

echo "current test time is ".(0-$last_status['step']).", max test time is $max_tcp_test_time";

echo "\n";

if($last_status['step'] > 0){

output_json(-1);

}else{

output_json($last_status['step']-1);

}

if($last_status['step'] <= -($max_tcp_test_time)){

echo "tcp test Reached the max test time: $max_tcp_test_time, will start swap ip process...";

echo "\n";

output_json(0);

}

}else{

echo "tcp test result is $tcp_test, server is running fine";

echo "\n";

}

Top Menu

Navigation

如何利用Vultr的API自动更换故障主机

切换流程

技术细节

公用函数

步骤零：获取主机列表及程序执行状态

使用状态码执行命令

步骤一：停用故障主机

步骤二：创建快照

步骤三：创建新服务器

步骤四：从快照恢复

步骤五：删除旧服务器

步骤六：更新WireGuard节点信息，Ping节点

循环步骤：检查TCP状态

There are no comments yet