php5-fpm chokes on missing /var/www

[CatchesAStar]

Binary package hint: php5-fpm

When php5-fpm is installed, it looks for a /var/www directory to generate its configuration.
However, php5-fpm is not only used for apache. Other webservers, such as Nginx do not use the /var/www folder as their root. As a result, php5-fpm stops during installation with

Setting up php5-fpm (5.3.5-1ubuntu1) ...
update-rc.d: warning: php5-fpm stop runlevel arguments (0 1 6) do not match LSB Default-Stop values (none)
 * Starting PHP5 FPM... Mar 08 20:14:16.908184 [WARNING] [pool www] pm.start_servers is not set. It's been set to 20.
Mar 08 20:14:16.908957 [ERROR] [pool www] the chdir path '/var/www' does not exist or is not a directory
Mar 08 20:14:16.908969 [ERROR] failed to post process the configuration

and the /var/www folder has to be created in order to install php5-fpm.

Maybe we can make a dialog or something else to let users choose the directory, or simply create the /var/www folder.

[Andreas Moog (ampelbein)]

Thanks for the report, that is already known at debian. See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=601243

[Launchpad Janitor (janitor)]

This bug was fixed in the package php5 - 5.3.5-1ubuntu2

---------------
php5 (5.3.5-1ubuntu2) natty; urgency=low

  * debian/php5-fpm.init: Dont start fpm if /var/www doesnt exist.
    (LP: #731572)
 -- Chuck Short <email address hidden> Fri, 11 Mar 2011 16:29:24 -0500

[David Strauss (davidstrauss)]

I have 5.3.5-1ubuntu4 installed, and this bug still affects me.

[David Strauss (davidstrauss)]

Steps to reproduce:

1. Install Natty server Alpha 3.
2. sudo apt-get install -y php5-fpm

[David Strauss (davidstrauss)]

Oh, I actually get a *warning* now, but the package still refuses to finish installation because of it. So, the original issue here isn't fixed.

[Ondřej Surý (ondrej)]

I'm here with David...

You didn't solve the original problem, but created more problems in one go. The init.d script really should fail if there's incorrect configuration, but this fix:

a) ignores the failure of missing /var/www (by returning
b) stops php5-fpm from running even if there is a correct configuration, but /var/www is missing

And thanks for fixing that only in Ubuntu, that's always helpful.

[Clint Byrum (clint-fewbar)]

Agreed, this isn't fixed as evidenced by the new bug report bug #754210

Ondrej, we generally fix things in Ubuntu first and then send the patch back to Debian later once we know the fix was effective for Ubuntu users.

[Ondřej Surý (ondrej)]

> Ondrej, we generally fix things in Ubuntu first and then send the patch
> back to Debian later once we know the fix was effective for Ubuntu users.

Maybe that's generally what you (should) do, but I don't remember a single patch provided by the Ubuntu to the Debian pkg-php team. I had to cherry-pick interesting stuff myself, and given that the Debian php team is undermanned for a long time, this isn't exactly a good base for fruitful cooperation...

I don't care for patches which apply only to Ubuntu, but where is the patch for:

LP: #739977
LP: #697181
LP: #611316
LP: #676672
LP: #638401
LP: #564920

Any many more... just picked few interesting pieces from Ubuntu changelog diff, or the difference Ubuntu and Debian debian/patches/series:

diff -pruN 5.3.5-1/debian/patches/series 5.3.5-1ubuntu6/debian/patches/series
--- 5.3.5-1/debian/patches/series 2011-04-01 20:47:19.000000000 +0100
+++ 5.3.5-1ubuntu6/debian/patches/series 2011-04-01 20:41:07.000000000 +0100
@@ -50,6 +50,8 @@ session_save_path.patch
 #581911_pdo_mysql_segfaults.patch
 php-fpm-man-section-and-cleanup.patch
 fpm-config.patch
+lp564920-fix-big-files.patch
+configure-as-needed.patch
 fix_crash_in__php_mssql_get_column_content_without_type.patch
 bug53070.patch
 fix-crash-with-entity-declarations-in-simplexml.patch
@@ -59,3 +61,5 @@ fix-memory-leak-inside-highlight_string.
 fix-segfault-in-pgsql_stmt_execute-when-postgres-is-down.patch
 fix-segfault-when-node-is-NULL-in-simplexml.patch
 fix-segfault-when-using-several-cloned-intl-objects.patch
+temporary-path-fixes-for-multiarch.patch
+backport-upstream-lp592442.patch

[Clint Byrum (clint-fewbar)]

Ondrej, I have to agree that we're not sending nearly enough patches back to you and your team, and you seem to be staying at least 1 step ahead of us at all times.

We'll keep working on it.. and thanks for the heads up on the patches we haven't even begun to send back.