Debian Bug report logs - #504132
libapache2-mod-fastcgi: fastcgi does not retry select() on EAGAIN

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Tobias Diedrich <ranma+debianbts@tdiedrich.de>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: libapache2-mod-fastcgi: fastcgi does not retry select() on EAGAIN

Date: Fri, 31 Oct 2008 21:37:00 +0100

Package: libapache2-mod-fastcgi
Version: 2.4.2-8
Severity: normal

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


I noticed that on our server fastcgi sometimes aborts because the select system
call was interrupted instead of retrying the system call:

[Fri Oct 31 20:14:47 2008] [error] [client xx.xxx.xx.xx] (4)Interrupted system call: FastCGI: comm with server "/xxx[...]/php5" aborted: select() failed, referer: http://www.aniki.info/Gott_Gauss

I googled this and found the following patch:
http://article.gmane.org/gmane.comp.web.fastcgi.devel/2514

However I think sleeping 1s before retrying is unnecessary (and having a
maximum amount of retries might be a bit paranoid), so I'm not saying this
patch should be applied as-is.

I looked at the libapache-mod-fastcgi-2.4.6 source package and AFAICS it also
has this bug.

Note that the select(2) manpage clearly states:
|Under Linux, select() may report a socket file descriptor as "ready for
|reading", while nevertheless a subsequent read blocks.  This could  for
|example  happen  when  data  has arrived but upon examination has wrong
|checksum and is discarded.  There may be other circumstances in which a
|file  descriptor is spuriously reported as ready.  Thus it may be safer
|to use O_NONBLOCK on sockets that should not block.

So this applies even though the linked mail only talks about this problem
occuring on AIX.

FYI I'm not submitting the bug directly from the server (because I get 'out of
memory' errors from reportbug due to a 32MB memory ulimit as user and I don't
want to reportbug as root), the server is running etch/stable, I have edited
the 'System Information' below accordingly.

- -- System Information:
Debian Release: etch/stable
Architecture: i686 (x86)

Kernel: Linux 2.6.25.4
Locale: LANG=en_IN, LC_CTYPE=en_IN (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iD8DBQFJC2xnzQZOfTz8JZwRAu+DAKCjdlP3YHXbAZ8J+odPMTiX5ShjHQCfZWzZ
fxDZ/qLegIKOqCAORtott7s=
=HwyU
-----END PGP SIGNATURE-----

Message #10 received at 504132@bugs.debian.org (full text, mbox, reply):

From: Tobias Diedrich <ranma+debianbts@tdiedrich.de>

To: Debian Bug Tracking System <504132@bugs.debian.org>

Subject: Re: libapache2-mod-fastcgi: fastcgi does not retry select() on EAGAIN

Date: Sun, 2 Nov 2008 19:35:32 +0100

Tobias Diedrich wrote:
> However I think sleeping 1s before retrying is unnecessary (and having a
> maximum amount of retries might be a bit paranoid), so I'm not saying this
> patch should be applied as-is.

I'm using the following modified patch now:

Index: libapache-mod-fastcgi-2.4.2/mod_fastcgi.c
===================================================================
--- libapache-mod-fastcgi-2.4.2.orig/mod_fastcgi.c	2008-11-02 16:42:49.000000000 +0000
+++ libapache-mod-fastcgi-2.4.2/mod_fastcgi.c	2008-11-02 16:50:46.000000000 +0000
@@ -2178,12 +2178,15 @@
         }
 
         /* wait on the socket */
-        select_status = ap_select(nfds, &read_set, &write_set, NULL, &timeout);
+        /* Interrupted system calls do happen now and then, so retry on EINTR */
+        do {
+            select_status = ap_select(nfds, &read_set, &write_set, NULL, &timeout);
+        } while (select_status < 0 && errno == EINTR);
 
         if (select_status < 0)
         {
             ap_log_rerror(FCGI_LOG_ERR_ERRNO, r, "FastCGI: comm with server "
-                "\"%s\" aborted: select() failed", fr->fs_path);
+                "\"%s\" aborted: select() failed: \"%s\"", fr->fs_path, strerror(errno));
             state = STATE_ERROR;
             break;
         }
@@ -2246,11 +2249,19 @@
             }
 
             rv = fcgi_buf_socket_recv(fr->serverInputBuffer, fr->fd);
+            /*
+             * select(2) states: Under Linux, select() may report a socket
+             * file descriptor as "ready for reading", while nevertheless a
+             * subsequent read blocks.
+             * Act as if the FD was not set if socket_recv returns EAGAIN.
+             */
+            if (rv < 0 && errno == EAGAIN)
+                break;
 
             if (rv < 0) 
             {
                 ap_log_rerror(FCGI_LOG_ERR, r, "FastCGI: comm with server "
-                    "\"%s\" aborted: read failed", fr->fs_path);
+                    "\"%s\" aborted: read failed: \"%s\"", fr->fs_path, strerror(errno));
                 state = STATE_ERROR;
                 break;
             }

-- 
Tobias						PGP: http://9ac7e0bc.uguu.de
このメールは十割再利用されたビットで作られています。

Message #15 received at 504132@bugs.debian.org (full text, mbox, reply):

From: CtRiX <ctrix+debian@navynet.it>

To: Debian Bug Tracking System <504132@bugs.debian.org>

Subject: libapache2-mod-fastcgi: Problem persists in lenny and SID

Date: Thu, 26 Feb 2009 17:55:25 +0100

Package: libapache2-mod-fastcgi
Version: 2.4.6-1
Followup-For: Bug #504132

The problem persists here.
It has been fixed in 2.4.7.
See http://www.fastcgi.com/archives/fastcgi-developers/2009-January/000156.html for some info.

Please backport on lenny and update at least sid.


-- System Information:
Debian Release: 5.0
  APT prefers stable
  APT policy: (500, 'stable')
Architecture: amd64 (x86_64)

Kernel: Linux 2.6.28.4-vserver-navynet (SMP w/4 CPU cores; PREEMPT)
Locale: LANG=it_IT.UTF-8, LC_CTYPE=it_IT.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

Versions of packages libapache2-mod-fastcgi depends on:
ii  apache2.2-common         2.2.9-10+lenny2 Apache HTTP Server common files
ii  libc6                    2.7-18          GNU C Library: Shared libraries

libapache2-mod-fastcgi recommends no packages.

libapache2-mod-fastcgi suggests no packages.

-- no debconf information

Message #20 received at 504132@bugs.debian.org (full text, mbox, reply):

From: Alexander Schories <alexander@schories.com>

To: 504132@bugs.debian.org

Subject: Please integrate the patch into the libapache2-mod-fastcgi Debian package as upstream has failed to release a corrected version since over 1 year

Date: Sun, 08 Nov 2009 23:45:58 +0100

Dear valued maintainers, Dear Tatsuki Sugiura,


first of all let me honestly thank you for maintaining this wonderfull
package for the Debian community.

Since November 13th, 2007 upstream has failed to release a newer or even
corrected fastcgi version (2.4.7).

Another year has passed since Tobias Diedrich reported the issue and
also kindly added a patch as a solution because - as we all know - the
expected upstream release that should correct this issue still hasn't
been released yet.

Many users (including myself) have to manually patch this wonderfull
package (successfully using this or a similar version of the patch) in
order to be able to use it without running into the described bug on
high load sites.

As Debian is most highly regarded for being a truly reliable operating
system, e.g. especially for sites with high traffic and true load, it
would be just perfect if this Debian package would include this
important fix by default.

Thank you once again for bringing this important package to the Debian
community!

Kind Regards


Alexander Schories
Tuebingen, Germany

Message #25 received at 504132@bugs.debian.org (full text, mbox, reply):

From: Alexander Schories <alexander@schories.com>

To: <504132@bugs.debian.org>, <592937@bugs.debian.org>, <thijs@debian.org>

Subject: Why libapache2-mod-fastcgi is still important

Date: Wed, 18 Aug 2010 15:45:14 +0200

Dear maintainers,


first of all let me thank you for maintaining this important web server
package for the Debian community.

libapache2-mod-fastcgi should be kept as libapache2-mod-fcgid is *NOT* a
proper replacement:

Today almost everybody uses PHP with a bytecode/opcode cacher (eg.
eAccelerator, xcache, apc,..). Opcode cachers, however, are sadly not able
to share the cache across FastCGI or FCGI processes. Means, for each new
PHP process the same opcode cache is built and filled up in the server’s
RAM. This is not only filling up the servers RAM with redundant data but is
actually ruining the performance of any opcode cacher and therefore all the
served web sites.

Luckily PHP is capable of playing “process manager” and a single PHP
process can spawn several children to handle requests. This way the parent
PHP process can instantiate the opcode cache and its children can share it.
Both libapache2-mod-fcgid and libapache2-mod-fastcgi can be told to limit
the number of PHP processes to 1 per user. The PHP process can then be told
how many children to spawn. 

But unfortunately libapache2-mod-fcgid will only send 1 single request per
process. The fact that PHP spawns its own children is completely ignored by
libapache2-mod-fcgid. So using libapache2-mod-fcgid we could only handle
one concurrent PHP request at a time. Obviously this is not good at all
since any longer running request would easily block multiple smaller
requests. 

libapache2-mod-fastcgi on the contrary will send multiple simultaneous
requests to a single PHP process if the PHP process itself has enough
children that can handle it. 

This is the reason we must keep libapache2-mod-fastcgi to achieve our goal
of one cache per user.


@ Tatsuki 

I am sure many users are as happy as i am that you maintain the package
for the community. However, almost 2 years have passed since bug report
#504132 that includes a working, common patch for the issue. Since then –
means over 2 years – you haven’t even replied once in that bug report
thread. If you are not capable or not willing to maintain this package any
longer i'll honestly be glad to help you.

Thank you very much once again


Alexander Schories
Tuebingen, Germany

Message #30 received at 504132@bugs.debian.org (full text, mbox, reply):

From: Sandro Tosi <morph@debian.org>

To: Tatsuki Sugiura <sugi@nemui.org>, Taku YASUI <tach@debian.or.jp>

Cc: 504132@bugs.debian.org, 509116@bugs.debian.org

Subject: What's the status of libapache-mod-fastcgi in Debian?

Date: Tue, 7 Jun 2011 12:09:56 +0200

Hi Tatsuki & Taku,
thanks for the work you did on libapache-mod-fastcgi! but recently, it
seems this package is lacking a bit of love :)

We'd like to use it at work, but we're facing several issues:

1. it doesn't retry to select() in case of EAGAIN, tracked at #504132[1]
2. mod_deflate and mod_fastcgid don't play together, tracked at #509116[2]
3. after the last upload, the package was not built on all the
architecture it was built before, as visible on the PTS[3] or directly
on the RT excuses[4]. and that prevents its migration to testing that
in return prevents its upload to backport.debian.org

[1] http://bugs.debian.org/504132
[2] http://bugs.debian.org/509116
[3] http://packages.qa.debian.org/liba/libapache-mod-fastcgi.html
[4] http://qa.debian.org/excuses.php?package=libapache-mod-fastcgi

1. and 2. have both patches and could be NMUed if needed (or you can
upload a new revision :)) while 3. is a bit more complicated.
Following [5] (keeping in mind [6]) you can read that "The inclusion
on the autobuild list can only be requested by the maintainer of the
package - we are not adding a package to be autobuild by requests of
anyone except the maintainer." so it's only you that can ask for such
autobuilding possibility (I didn't ask if this requirement can be
relaxed). Could you please follow those steps and ask for
libapache-mod-fastcgi autobuilding? another solution could be a
partial (arch) removal from unstable of the outdated archs, but I'd
like to use it as a last resort solution.

[5] http://lists.debian.org/debian-devel-announce/2006/11/msg00012.html
[6] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=612563#10

Please note that this is a work project, so I'm feeling a little bit
of pressure and I cannot wait that long :) I can NMU or prepare an
internal package, but I'd like much more to find a distribution
solution so that others can benefit too. Please let me know as soon as
you can, but it might be that in about a week I'll start to move this
forward.

Thanks a lot in advance,
-- 
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

Message #35 received at 504132@bugs.debian.org (full text, mbox, reply):

From: serh.yakymenko@gmail.com

To: linux.debian.bugs.dist@googlegroups.com

Cc: Tobias Diedrich <ranma+debianbts@tdiedrich.de>, 504132@bugs.debian.org

Subject: Re: Bug#504132: libapache2-mod-fastcgi: fastcgi does not retry select() on EAGAIN

Date: Sun, 9 Dec 2012 01:43:17 -0800 (PST)

Hi to everybody, long time ago this post was published but it works!!!

Using Freebsd 7.3 apache 2.2 php 5.4.7 php-fpm mod_fastcgi 2.4.6 and got select() failed every time my long script started.

To apply it:

1. Make deinstall to module without clean.
2. Edit source file Tobias says in port www/mod_fastcgi/...
3. Run command to rebuild module noted in INSTALL ( apxs -o mod_fastcgi.so -c *.c ).
4. Make install.
5. Restart apache.

Dont know who is author of module, but this change should be commited in the core of module.

Many thanks to athor of patch.
Many thanks to mod_fastcgi athor.
Many thanks to apache and google community.

See you. 

Неділя, 2 листопада 2008 р. 20:40:08 UTC+2 користувач Tobias Diedrich написав:
> Tobias Diedrich wrote:
> > However I think sleeping 1s before retrying is unnecessary (and having a
> > maximum amount of retries might be a bit paranoid), so I'm not saying this
> > patch should be applied as-is.
> 
> I'm using the following modified patch now:
> 
> Index: libapache-mod-fastcgi-2.4.2/mod_fastcgi.c
> ===================================================================
> --- libapache-mod-fastcgi-2.4.2.orig/mod_fastcgi.c	2008-11-02 16:42:49.000000000 +0000
> +++ libapache-mod-fastcgi-2.4.2/mod_fastcgi.c	2008-11-02 16:50:46.000000000 +0000
> @@ -2178,12 +2178,15 @@
>          }
>  
>          /* wait on the socket */
> -        select_status = ap_select(nfds, &read_set, &write_set, NULL, &timeout);
> +        /* Interrupted system calls do happen now and then, so retry on EINTR */
> +        do {
> +            select_status = ap_select(nfds, &read_set, &write_set, NULL, &timeout);
> +        } while (select_status < 0 && errno == EINTR);
>  
>          if (select_status < 0)
>          {
>              ap_log_rerror(FCGI_LOG_ERR_ERRNO, r, "FastCGI: comm with server "
> -                "\"%s\" aborted: select() failed", fr->fs_path);
> +                "\"%s\" aborted: select() failed: \"%s\"", fr->fs_path, strerror(errno));
>              state = STATE_ERROR;
>              break;
>          }
> @@ -2246,11 +2249,19 @@
>              }
>  
>              rv = fcgi_buf_socket_recv(fr->serverInputBuffer, fr->fd);
> +            /*
> +             * select(2) states: Under Linux, select() may report a socket
> +             * file descriptor as "ready for reading", while nevertheless a
> +             * subsequent read blocks.
> +             * Act as if the FD was not set if socket_recv returns EAGAIN.
> +             */
> +            if (rv < 0 && errno == EAGAIN)
> +                break;
>  
>              if (rv < 0) 
>              {
>                  ap_log_rerror(FCGI_LOG_ERR, r, "FastCGI: comm with server "
> -                    "\"%s\" aborted: read failed", fr->fs_path);
> +                    "\"%s\" aborted: read failed: \"%s\"", fr->fs_path, strerror(errno));
>                  state = STATE_ERROR;
>                  break;
>              }
> 
> -- 
> Tobias						PGP: http://9ac7e0bc.uguu.de
> このメールは十割再利用されたビットで作られています。
> 
> 
> 
> -- 
> To UNSUBSCRIBE, email to debian-bugs-dist-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Message #40 received at 504132@bugs.debian.org (full text, mbox, reply):

From: Pavel Odintsov <pavel.odintsov@gmail.com>

To: 504132@bugs.debian.org

Subject: Patch for upgrading mod_fastcgi from 2.6.4 to SNAP 2.4.7~0910052141

Date: Tue, 5 Mar 2013 12:41:33 +0400

Hello,

I created patch for upgrade mod_fastcgi up to SNAP 2.4.7~0910052141
version from 2.6.4 to SNAP 2.4.7~0910052141 using sources from
http://www.fastcgi.com/dist/.

U can find standard patch between versions here (I'm excluded debian
folder from upstream completely):
https://code.google.com/p/fastvps/source/browse/trunk/patches/mod_fastcgi/upgrade_mod_fastcgi_to_snap_2_4_7_0910052141.patch

I created dpatch also
(https://code.google.com/p/fastvps/source/browse/trunk/patches/mod_fastcgi/upgrade_mod_fastcgi_to_snap_2_4_7_0910052141.dpatch),
u can patch standard squeeze buggy  package very simple:
cd /usr/src
apt-get install -y dpkg-dev devscripts build-essential fakeroot
apt-get build-dep  libapache2-mod-fastcgi
apt-get source  libapache2-mod-fastcgi
cd libapache-mod-fastcgi-2.4.6/debian/patches
wget https://fastvps.googlecode.com/svn/trunk/patches/mod_fastcgi/upgrade_mod_fastcgi_to_snap_2_4_7_0910052141.dpatch
echo "upgrade_mod_fastcgi_to_snap_2_4_7_0910052141.dpatch" > 00list
cd ..
cd ..
debuild -us -uc

Also u can download my precreated binary package:
https://fastvps.googlecode.com/svn/trunk/patches/mod_fastcgi/upgrade_mod_fastcgi_to_snap_2_4_7_0910052141.dpatch

--
Pavel Odintsov

Message #45 received at 504132-done@bugs.debian.org (full text, mbox, reply):

From: Debian FTP Masters <ftpmaster@ftp-master.debian.org>

To: 504132-done@bugs.debian.org,509116-done@bugs.debian.org,580574-done@bugs.debian.org,592937-done@bugs.debian.org,650953-done@bugs.debian.org,656425-done@bugs.debian.org,664348-done@bugs.debian.org,701118-done@bugs.debian.org,730208-done@bugs.debian.org,754444-done@bugs.debian.org,803063-done@bugs.debian.org,

Cc: libapache-mod-fastcgi@packages.debian.org, libapache-mod-fastcgi@packages.qa.debian.org

Subject: Bug#835374: Removed package(s) from unstable

Date: Fri, 14 Oct 2016 11:51:37 +0000

Version: 2.4.7~0910052141-1.2+rm

Dear submitter,

as the package libapache-mod-fastcgi has just been removed from the Debian archive
unstable we hereby close the associated bug reports.  We are sorry
that we couldn't deal with your issue properly.

For details on the removal, please see https://bugs.debian.org/835374

The version of this package that was in Debian prior to this removal
can still be found using http://snapshot.debian.org/.

This message was generated automatically; if you believe that there is
a problem with it please contact the archive administrators by mailing
ftpmaster@ftp-master.debian.org.

Debian distribution maintenance software
pp.
Scott Kitterman (the ftpmaster behind the curtain)

Send a report that this bug log contains spam.