[FrameworkBundle] Additional helper commands to control PHP's built-in web server #11311

xabbuh · 2014-07-05T11:01:42Z

Q	A
Bug fix?	no
New feature?	yes
BC breaks?	no
Deprecations?	no
Tests pass?	yes
Fixed tickets	#10827
License	MIT
Doc PR	symfony/symfony-docs#4005

Basically, both the server:status and server:stop wouldn't be really reliable if you had stopped the web server by, for example, killing the process. But honestly I don't know how to platform-independently determine if a process is still running given its PID. Maybe such a way could be a good improvement for the Process component.

romainneutron · 2014-07-05T11:18:42Z

src/Symfony/Component/Process/ProcessBuilder.php

@@ -256,7 +256,7 @@ public function getProcess()
        $options = $this->options;

        $arguments = array_merge($this->prefix, $this->arguments);
-        $script = implode(' ', array_map(array(__NAMESPACE__.'\\ProcessUtils', 'escapeArgument'), $arguments));
+        $script = $arguments[0].implode(' ', array_map(array(__NAMESPACE__.'\\ProcessUtils', 'escapeArgument'), array_slice($arguments, 1)));


please revert this change

Without this change I wasn't able to prefix the PHP_BINARY call with exec. Do you have any idea how to handle it then?

@romainneutron To be more precise: The command that should be executed leading to a "Command not found error" actually is something like:

'exec' '/usr/bin/php' ...

Either we could implement a new feature in the process builder, either you can build the command manually instead of using the ProcessBuilder, but you can not decide to unescape the first argument.

Of course, you're right. I first thought that it doesn't make sense to escape the command name. But that's obviously not always true. What do you think about adding a method like useExec to the ProcessBuilder class?

That would be indeed a nice feature. Please open a new PR for this feature as it would be easier to review this new feature in a separate thread

I created a pull request for this (see #11335).

xabbuh · 2014-07-06T14:41:24Z

Both the server:status and server:stop commands have to rely on the fact that the server:status command terminates the proper process. The usage of exec is required for this to work as expected. Thus, #11335 or a similar solution needs to be merged first (currently, this change is based on the modifications made for #11335).

GromNaN · 2014-07-07T20:11:12Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStatusCommand.php

+    {
+        $address = $input->getArgument('address');
+
+        if (file_exists($this->getLockFile($address))) {


The server process could have crashed. You could check if the server sends HTTP responses.

Good idea. That could be done relatively easy.

The lock file is now removed if the server process terminated before.

xabbuh · 2014-07-08T18:33:52Z

I modified this a bit. The pull request is no longer based off #11335. Instead the script that is executed by the start:start command is built manually which was way easier than I had expected.

xabbuh · 2014-08-15T17:43:33Z

Any thoughts on this?

fabpot · 2014-08-28T03:31:50Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStartCommand.php

+        } elseif ($pid > 0) {
+            $output->writeln(sprintf(
+                '<info>Server started successfully</info>',
+                $pid


This var is unused in the string.

Removed it.

fabpot · 2014-08-28T06:45:39Z

I've not tried it but it looks good to me.

fabpot · 2014-09-02T14:07:33Z

@romainneutron should give you some feedback soon...

romainneutron · 2014-09-03T08:31:39Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStatusCommand.php

+    {
+        list($hostname, $port) = explode(':', $address);
+
+        return @fsockopen($hostname, $port, $errno, $errstr, 3000) !== false;


3000 seconds is very long timeout, I rather suggest a short one, 1 second should be enough.

Moreover, this method should return a boolean and close the socket resource once it has been created

romainneutron · 2014-09-03T08:34:30Z

It would be nice to add correct exitcodes (0 on success, other codes in case of failures)

xabbuh · 2014-09-03T08:50:06Z

@romainneutron I added 1 where necessary. 0 is the default behaviour. Do you think it makes sense to return 0 explicitly?

romainneutron · 2014-09-03T08:53:38Z

No need to return 0 in case the command was successful

lyrixx · 2014-09-03T08:54:20Z

The lock management could/should use #10475 (but it's not yet merged)

xabbuh · 2014-09-03T08:54:33Z

You can then have another look.

romainneutron · 2014-09-03T08:55:42Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStatusCommand.php

+    {
+        list($hostname, $port) = explode(':', $address);
+        $fp = @fsockopen($hostname, $port, $errno, $errstr, 1);
+        fclose($fp);


If it failed, this call would fail. You should use instead:

if (false !== $fp = @fsockopen($hostname, $port, $errno, $errstr, 1)) { fclose($fp); return true; } return false;

Oh yes, that's true.

romainneutron · 2014-09-03T09:05:48Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStartCommand.php

+            $process->start();
+            $lockFile = $this->getLockFile($input->getArgument('address'));
+            touch($lockFile);
+


Let's add a check: If the server could not be started, let's display an error

if (!$process->isRunning()) { $output->writeln('<error>Unable to start the server process</error>'); unlink($lockfile); return 1; }

romainneutron · 2014-09-03T09:10:20Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStatusCommand.php

+
+    private function isServerRunning($address)
+    {
+        list($hostname, $port) = explode(':', $address);


this would trigger an error in case the address is not in the right format (for example randomstring would trigger an Undefined index notice

To avoid that, you can use:

explode(':', $address) + array(8000);

or better, ensure the passed argument matches a regular expression

Not really, there wouldn't be any server running with this address. So, there is no lock file and isServerRunning() would never be called (see line 48).

xabbuh · 2014-09-03T17:28:40Z

I addressed your other suggestions.

romainneutron · 2014-09-09T08:11:43Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStartCommand.php

+            return 1;
+        } elseif ($pid > 0) {
+            $output->writeln(sprintf(
+                '<info>Server started successfully</info>',


$pid argument does not seem to be used with sprintf

romainneutron · 2014-09-09T08:21:45Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStartCommand.php

+                $pid
+            ));
+        } else {
+            $process = $this->createServerProcess(


It should be good to add a call to posix_setsid() here to avoid zombie process.

if (posix_setsid() < 0) { return 1; }

xabbuh · 2014-09-09T09:20:01Z

Thanks for the feedback.

romainneutron · 2014-09-09T09:23:38Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStartCommand.php

+            $output->writeln('<info>Server started successfully</info>');
+        } else {
+            if (posix_setsid() < 0) {
+                return 1;


might be nice to display a nice message in case it happens like Unable to set the child process as session leader

romainneutron · 2014-09-09T09:23:59Z

I'm 👍 on this one (once my last comment would be addressed)

stof · 2014-09-09T09:34:31Z

src/Symfony/Bundle/FrameworkBundle/Command/ServerStartCommand.php

+            $output->writeln('<error>Unable to start the server process</error>');
+
+            return 1;
+        } elseif ($pid > 0) {


just if as the previous if returns

xabbuh · 2014-09-21T18:38:53Z

Any other opinion on this?

fabpot · 2014-09-21T18:52:15Z

👍

fabpot · 2014-09-22T12:30:18Z

Thank you @xabbuh.

…l PHP's built-in web server (xabbuh) This PR was merged into the 2.6-dev branch. Discussion ---------- [FrameworkBundle] Additional helper commands to control PHP's built-in web server | Q | A | ------------- | --- | Bug fix? | no | New feature? | yes | BC breaks? | no | Deprecations? | no | Tests pass? | yes | Fixed tickets | #10827 | License | MIT | Doc PR | symfony/symfony-docs#4005 Basically, both the ``server:status`` and ``server:stop`` wouldn't be really reliable if you had stopped the web server by, for example, killing the process. But honestly I don't know how to platform-independently determine if a process is still running given its PID. Maybe such a way could be a good improvement for the Process component. Commits ------- b601454 new helper commands for PHP's built-in server

…mands (xabbuh) This PR was merged into the 2.6-dev branch. Discussion ---------- [FrameworkBundle] add changelog entry for web server commands | Q | A | ------------- | --- | Bug fix? | no | New feature? | no | BC breaks? | no | Deprecations? | no | Tests pass? | yes | Fixed tickets | | License | MIT | Doc PR | I just noticed that I forgot to add a changelog entry in #11311. Commits ------- 292088f add changelog entry for web server commands

…ilt-in web server in the background (xabbuh) This PR was merged into the master branch. Discussion ---------- [Cookbook][Web server] description for running PHP's built-in web server in the background | Q | A | ------------- | --- | Doc fix? | no | New docs? | yes (symfony/symfony#11311) | Applies to | 2.6+ | Fixed tickets | Commits ------- 3caec21 description for running PHP's built-in web server in the background

romainneutron reviewed Jul 5, 2014
View reviewed changes

romainneutron mentioned this pull request Jul 5, 2014

[Filesystem] Added a LockHandler #10475

Merged

xabbuh mentioned this pull request Jul 7, 2014

[Cookbook][Web server] description for running PHP's built-in web server in the background symfony/symfony-docs#4005

Merged

GromNaN reviewed Jul 7, 2014
View reviewed changes

romainneutron added the FrameworkBundle label Jul 12, 2014

fabpot reviewed Aug 28, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch from fecfe47 to 2b27268 Compare August 28, 2014 06:03

romainneutron reviewed Sep 3, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch 2 times, most recently from 4fbf128 to 459a97e Compare September 3, 2014 08:48

xabbuh force-pushed the issue-10827 branch from 459a97e to d56179b Compare September 3, 2014 08:51

romainneutron reviewed Sep 3, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch 2 times, most recently from bde1b5e to e796bac Compare September 3, 2014 08:59

romainneutron reviewed Sep 3, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch from e796bac to cc197be Compare September 3, 2014 17:28

romainneutron reviewed Sep 9, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch 2 times, most recently from 28afe29 to 74c993c Compare September 9, 2014 08:16

romainneutron reviewed Sep 9, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch from 74c993c to 5af7a1a Compare September 9, 2014 08:35

romainneutron reviewed Sep 9, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch from 5af7a1a to 88c869b Compare September 9, 2014 09:33

stof reviewed Sep 9, 2014
View reviewed changes

xabbuh force-pushed the issue-10827 branch 2 times, most recently from 8464681 to bd58bcb Compare September 9, 2014 09:51

new helper commands for PHP's built-in server

b601454

xabbuh force-pushed the issue-10827 branch from bd58bcb to b601454 Compare September 9, 2014 09:52

fabpot merged commit b601454 into symfony:master Sep 22, 2014

xabbuh deleted the issue-10827 branch September 22, 2014 12:45

xabbuh mentioned this pull request Oct 9, 2014

[FrameworkBundle] add changelog entry for web server commands #12184

Merged

xabbuh mentioned this pull request Feb 1, 2016

introduce new helper commands for PHP's built-in server #17635

Closed

[FrameworkBundle] Additional helper commands to control PHP's built-in web server #11311

[FrameworkBundle] Additional helper commands to control PHP's built-in web server #11311

Conversation

xabbuh commented Jul 5, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

xabbuh commented Jul 6, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

xabbuh commented Jul 8, 2014

xabbuh commented Aug 15, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fabpot commented Aug 28, 2014

fabpot commented Sep 2, 2014

Choose a reason for hiding this comment

romainneutron commented Sep 3, 2014

xabbuh commented Sep 3, 2014

romainneutron commented Sep 3, 2014

lyrixx commented Sep 3, 2014

xabbuh commented Sep 3, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

xabbuh commented Sep 3, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

xabbuh commented Sep 9, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

romainneutron commented Sep 9, 2014

Choose a reason for hiding this comment

xabbuh commented Sep 21, 2014

fabpot commented Sep 21, 2014

fabpot commented Sep 22, 2014