Can't get scripts/getvaliddaemons.php to give the TwitterBridge daemons

mvdan

Hello guys!

I've recently installed a fresh 1.1.0-alpha1 statusnet instance on my server, http://status.cat/. It works perfectly under PHP-FPM 5.4 and a NGINX (engine-x) web server, the web interface doesn't give any errors at all. I've also activated a few plug-ins through config.php, as follows:

-- config.php --

<?php<br />if (!defined('STATUSNET') && !defined('LACONICA')) { exit(1); }

$config['site']['name'] = 'Status.cat';
$config['site']['server'] = 'status.cat';
$config['site']['path'] = false;

$config['site']['logdebug'] = false;
$config['site']['logfile'] = '/var/log/statusnet/statuscat.log';

$config['site']['theme'] = 'neo-light';
$config['site']['fancy'] = true;

$config['daemon']['user'] = "http";
$config['daemon']['group'] = "http";
$config['daemon']['piddir'] = "/var/run/statusnet/";

$config['db']['database'] = 'mysqli://___:___@localhost/___';
$config['db']['type'] = 'mysql';
$config['queue']['enabled'] = true;

$config['site']['profile'] = 'public';
$config['public']['localonly'] = false;
$config['sms']['enabled'] = false;

$config['twitter']['enabled'] = true;
$config['twitterimport']['enabled'] = true;
$config['integration']['source'] = 'statuscat';
$config['admin']['panels'][] = 'twitter';

addPlugin('InProcessCache');
addPlugin('Memcached');
addPlugin('Gravatar');
addPlugin('InfiniteScroll');
addPlugin('WikiHashtags');
addPlugin('SubMirror');
addPlugin('LinkPreview');
addPlugin('Blacklist');
addPlugin('TwitterBridge', array(
'consumer_key' => '_______________',
'consumer_secret' => '__________________________________'));
addPlugin('recaptcha', array('private_key' => '____________________',
'public_key' => '___________________________________',
'display_errors' => true));
addPlugin('Autocomplete');
addPlugin('TabFocus');

--

They all work. The problem I'm having with TwitterBridge is that for some reason, "sudo -u http php scripts/getvaliddaemons.php" only gives imdaemon.php and queuedaemon.php, making scripts/startdaemons.sh forget about the twitter ones. As an ugly work-around to this problem, I made a backup of getvaliddaemons.php and added manually the three under plugins/TwitterBridge/daemons/, but they won't get daemonized. No errors are thrown in the shell.

Any idea as to why would this happen? TwitterBridge was enabled after I switched to 5.4, so I don't know if that's the problem. Although since everything else seems to work flawlessly, I suppose there's something I'm doing wrong. Posts from status.cat get forwarded correctly to twitter, by the way. The app in twitter has full read/write access.

Thanks!

P.S. I don't want to start new threads, but here are two quick questions:
1. My linode machine runs their kernel, which doesn't include /dev/urandom, thus bringing OpenID unusable. I tried using /dev/random and even /dev/null in its php file, but to no luck. Any workaround available?
2. The InifiniteScroll plugin works, but the icon won't show up. This error is thrown:
"GET /mvdan/allplugins/InfiniteScroll/ajax-loader.gif HTTP/1.1" 404 1532 "http://status.cat/"
Why is it looking under /mvdan/allplugins instead of /plugins? I'm confused.

lnxwalt

@mvdan

I don't know anyone who is using PHP 5.4 yet (except possibly @jpope). He's also running the alpha version of SN. Hopefully, he or someone else who is using these versions knows what is going on.

One question: What distro are you using?

jpope

For reference, here is one thread that @mvdan and I have had going concerning this. link

I am currently at a loss as to why getvaliddaemons.php doesn't pick up the twitter daemons. I do only have two twitter related daemons that run on my instance,

./plugins/TwitterBridge/daemons/synctwitterfriends.php 
./plugins/TwitterBridge/daemons/twitterstatusfetcher.php

The twitterdaemon.php script doesn't daemonize for me at all, maybe try removing that one from getvaliddaemons.php and see if that helps.

I just looked at status.cat and it looks that the InfiniteScroll is working (ajax-loader.gif loads for me), did you fix this or do you still get the error? The times that I've had that plugin running, my httpd error log gets hammered with

Undefined property: InfiniteScrollPlugin::$on_next_only...

errors. Instead of the InfiniteScroll, I've just increased the number of notices that show per page.

mvdan

@lnxwalt I was stupid enough not to try statusnet-1.1.0-alpha1 before the switch to 5.4. I'm using Arch Linux on my Linode, with linux 3.2 (their own kernel).

@Hmm, at this time your link seems to be down. Nevertheless, all we were talking about is mentioned in this thread.
[EDIT] Hah, took me so long to write this comment that it works now.

About the ajax plugin thing... Darn it. It is not the server, it's my browser. Now that I look at it from Firefox, it works. It doesn't from dwb. Definitely not statusnet's issue :)

And just to make sure there was nothing I messed up, yesterday night I deleted both database and directory, restarted mysql and flushed memcached. Fresh install once again, copying back config.php after /install.php was run. But the same issues are not gone.

I'll just go ahead and post here all my poking around with php:

# Getting daemons
[root@linode] [/srv/http/mvdan/statuscat]
# sudo -u http scripts/checkschema.php
[root@linode] [/srv/http/mvdan/statuscat]
# sudo -u http scripts/getvaliddaemons.php
/srv/http/mvdan/statuscat/scripts/queuedaemon.php
/srv/http/mvdan/statuscat/scripts/imdaemon.php

# Same as running "sudo -u http scripts/startdaemons.sh"
[root@linode] [/srv/http/mvdan/statuscat]
# sudo -u http php scripts/queuedaemon.php && php scripts/imdaemon.php
[root@linode] [/srv/http/mvdan/statuscat]
# ps aux | grep -i php
root 19980 0.0 1.0 228772 5448 ? Ss May20 0:01 php-fpm: master process (/etc/php/php-fpm.conf)
http 19981 0.0 8.3 242820 41584 ? S May20 0:46 php-fpm: pool www
http 19982 0.0 7.3 238956 36844 ? S May20 0:45 php-fpm: pool www
http 24689 0.2 3.2 211740 16188 pts/0 S 12:11 0:00 php scripts/queuedaemon.php
http 24690 0.0 3.2 212020 16456 pts/0 S 12:11 0:00 php scripts/queuedaemon.php
http 24694 0.0 3.1 211236 15644 pts/0 S 12:11 0:00 php scripts/imdaemon.php
http 24695 0.0 3.1 211512 15948 pts/0 S 12:11 0:00 php scripts/imdaemon.php
http 24697 0.0 3.2 212020 16452 pts/0 S 12:11 0:00 php scripts/queuedaemon.php
http 24699 0.0 3.2 212020 16452 pts/0 S 12:11 0:00 php scripts/queuedaemon.php
http 24701 0.0 3.2 212020 16452 pts/0 S 12:11 0:00 php scripts/queuedaemon.php
root 24717 0.0 0.2 8652 1076 pts/0 R+ 12:11 0:00 grep --color=auto -i php
http 24922 0.0 6.8 238460 34328 ? S May20 0:22 php-fpm: pool www

# And still stuck at those errors. Before I said there were no errors, my apologies. I forgot there are. php-pear is installed from [extra] and configured as default. Also, mysqli_reconnect is on. Maybe this thread is closely related to this one: http://forum.status.net/discussion/1932/twitterbridge-and-pear-db-errors
[root@linode] [/srv/http/mvdan/statuscat]
# sudo -u http php plugins/TwitterBridge/daemons/twitterstatusfetcher.php && php plugins/TwitterBridge/daemons/synctwitterfriends.php
PHP Fatal error: Uncaught PEAR_Exception: DB Error: no database selected in unknown on line unknown
DB_Error: DB Error: no database selected in unknown on line unknown
#0 [internal function]: PEAR_ErrorToPEAR_Exception(Object(DB_Error))
#1 /srv/http/mvdan/statuscat/extlib/PEAR.php(890): call_user_func('PEAR_ErrorToPEA...', Object(DB_Error))
#2 /srv/http/mvdan/statuscat/extlib/DB.php(966): PEAR_Error->PEAR_Error('DB Error: no da...', -14, 16, 'PEAR_ErrorToPEA...', 'SELECT * ? FROM...')
#3 /srv/http/mvdan/statuscat/extlib/PEAR.php(531): DB_Error->DB_Error(-14, 16, 'PEAR_ErrorToPEA...', 'SELECT * ? FROM...')
#4 /srv/http/mvdan/statuscat/extlib/DB/common.php(1908): PEAR->raiseError(NULL, -14, NULL, NULL, 'SELECT * ? FROM...', 'DB_Error', true)
#5 /srv/http/mvdan/statuscat/extlib/DB/mysqli.php(928): DB_common->raiseError(-14, NULL, NULL, NULL, '2006 ** MySQL s...')
#6 /srv/http/mvdan/statuscat/extlib/DB/mysqli.php(387): DB_mysqli->mysqliRaiseError(-14)
#7 /srv/http/mvdan/statuscat/extlib/DB/common.php(1216): DB_mysqli->simpleQuer in /srv/http/mvdan/statuscat/lib/framework.php on line 163
PHP Fatal error: Uncaught PEAR_Exception: DB Error: no database selected in unknown on line unknown
DB_Error: DB Error: no database selected in unknown on line unknown
#0 [internal function]: PEAR_ErrorToPEAR_Exception(Object(DB_Error))
#1 /srv/http/mvdan/statuscat/extlib/PEAR.php(890): call_user_func('PEAR_ErrorToPEA...', Object(DB_Error))
#2 /srv/http/mvdan/statuscat/extlib/DB.php(966): PEAR_Error->PEAR_Error('DB Error: no da...', -14, 16, 'PEAR_ErrorToPEA...', 'SELECT * ? FROM...')
#3 /srv/http/mvdan/statuscat/extlib/PEAR.php(531): DB_Error->DB_Error(-14, 16, 'PEAR_ErrorToPEA...', 'SELECT * ? FROM...')
#4 /srv/http/mvdan/statuscat/extlib/DB/common.php(1908): PEAR->raiseError(NULL, -14, NULL, NULL, 'SELECT * ? FROM...', 'DB_Error', true)
#5 /srv/http/mvdan/statuscat/extlib/DB/mysqli.php(928): DB_common->raiseError(-14, NULL, NULL, NULL, '2006 ** MySQL s...')
#6 /srv/http/mvdan/statuscat/extlib/DB/mysqli.php(387): DB_mysqli->mysqliRaiseError(-14)
#7 /srv/http/mvdan/statuscat/extlib/DB/common.php(1216): DB_mysqli->simpleQuer in /srv/http/mvdan/statuscat/lib/framework.php on line 163

They don't seem to have found a solution for it. @lnxwalt, is that problem happening with 5.3? But the twitterbridge daemons do appear in scripts/getvaliddaemons.php, right?

@jpope it might be helpful to roll back to 5.3 for a moment and see if it behaves the same way. Any idea as to how to do that?

Cheers!

jpope

@mvdan I'd certainly give a rollback to 5.3 a try. I attempted the 5.4 upgrade, noticed too many errors (Seems the XMPP plugin had some issues as well) and decided to just rollback for now and am currently running 5.3.11-1. I'm planning on setting up a dev webserver on my desktop to test things out sometime soon, just got distracted with my current site redesign... ;)

If you previously had 5.3 installed, you'll most likely have the old packages in /var/cache/pacman/pkg/. If so, you can downgrade by issuing a

#  pacman -U /var/cache/pacman/pkg/packagename.pkg.tar.xz

for each package you need to downgrade. I think you can list all the packages in one command if you want, it'll probably complain about dependencies if you don't... If you don't have the old packages in your cache, you should be able to find the old packages via the ARM (Arch Rollback Machine). Just wget the files you need and pacman -U to install. Also, if you have 'php-memcache', the most current version has issues as well with 5.3, I have 3.0.6-2 installed and working properly.

Also, after you have finished, you may want to add each of these package names to the 'IgnorePkg' line in /etc/pacman.conf so that on your next pacman -Syu, it doesn't try to update these. Otherwise, if you use yaourt as opposed to pacman directly, you can manually edit your update list after running 'yaourt -Syu' before actually updating.

Also, check the Arch wiki for more on downgrading, link

mvdan

I can't say I tried. Went back to 5.3.12:
# php --version
PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php/modules/xsl.so' - /usr/lib/php/modules/xsl.so: undefined symbol: zend_new_interned_string in Unknown on line 0
PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php/modules/apc.so' - /usr/lib/php/modules/apc.so: undefined symbol: zend_new_interned_string in Unknown on line 0
PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php/modules/memcached.so' - libmemcached.so.9: cannot open shared object file: No such file or directory in Unknown on line 0
PHP 5.3.12 with Suhosin-Patch (cli) (built: May 3 2012 18:18:26)
Copyright (c) 1997-2012 The PHP Group
Zend Engine v2.3.0, Copyright (c) 1998-2012 Zend Technologies

I panicked, and so rolled back to 5.4. Cloud you please paste "pacman -Ss php | grep installed" so I know which versions of which packages do I need? It seems like I'm missing something.

tonybaldwin

Ha! Well this cleared up why I was having difficulty with openid on my SN and friendica installations on my linode.

As you've noted, I get similar PearDB errors, and my twitterfetcher randomly dies and has to be restarted.
I added the two twitter scripts to the stopdaemon.sh, and now have cron run that and the startdaemons.sh early in the morning, just before I get up, so I can get the news from the twitter when I first get up...Not really a valid solution, of course. I wish more news outlets used SN!

Elsewhere it was suggested to put mysqli.reconnect = On in php.ini (in either/both /etc/php5/cli or /etc/php5/apache2), but I already have that on, and it hasn't solved the problem.
I might just be overworking mysql, since I have a few other apps using it (friendica being the most demanding other than SN), and only a 1gb of ram.

tonybaldwin

Awesome! Found El Periodico, a valued news source, on your status.cat

tonybaldwin

Oh, although I generally read it in Castellano (Catalan is not in my language list, although I can make some sense of it, of course).

mvdan

@tonybaldwin Yes, the kernel from linode is rather old and doesn't include urandom. Which is a pity. I would use my own kernel if it were as easy as a simple package install, but since it needs special building I think I'll leave it like that for now. OpenID isn't a priority.

And sorry, but where do you see any other php.ini? I only have one in /etc/php, there's no such /etc/php5 (I use nginx, no apache).

Oh, and as far as the El Periodico thing, I intend status.cat to be a social network for catalan people, thus sourcing some of the most popular catalan newspapers. El periodico has an rss feed in spanish, if you're interested.

[EDIT] Cool, I'm not the only one who is having problems with scripts/getvaliddaemons.php. Which PHP version are you running? If you're using 5.3, we can definitely say PHP 5.4 *is not* the cause of our issues. Then what might be? A bug in 1.1.0-alpha1? The absence of /dev/urandom? Some problem with mysql/memcached?

jpope

@mvdan Ah, sorry for the delay in my response. The output of pacman -Ss php ... is here: http://bin.jpope.org/rohupabuha

As far as I know the only php.ini resides at /etc/php/php.ini.

mvdan

Thanks! I'll be running a test ASAP when I get back home :)

jpope

No problem. It also may be worth noting that some of those php packages I have installed were not necessarily installed for SN but for one of the other dozen things I have running on my server. ;)

mvdan

@jpope Stupid me, it's not pacman -Ss I want. It's pacman -Qs. What you gave me is the ones from the updated repositories ;) Could you please send again? Sorry about that.

tonybaldwin

I follow the castellano ElPeriodico elsewhere, indeed.

My linode (debian stable) has kernel 3.0

uname -a gives
Linux vulcan 3.0.18-linode43

tonybaldwin

Oh, I have two php.ini.
one in /etc/php5/apache2/php.ini and the other in /etc/php5/cli/php.ini

this is debian stable.

lnxwalt

@mvdan On my Linode, 'cat /dev/urandom' works. Are you sure your kernel is missing urandom?

mvdan

# uname -a
Linux linode 3.2.1-x86_64-linode23 #1 SMP Mon Jan 23 13:37:32 EST 2012 x86_64 GNU/Linux

Now that's strange. Urandom *is* there. Hmm. Either I messed up earlier, or they updated the 3.2.1 kernel. Anyhoo, solved. I'll try and use openid again with /dev/urandom and see if it works.
[EDIT] It doesn't work. OpenID deactivated again.

@tonybaldwin I used to have /etc/php5 in debian as well. In arch, you get /etc/php/php.ini, no cli/ folder. And since Apache is not installed, no apache2 either. Though it's the same.

mvdan

(this comment was a wrong approach)

mvdan

Guys,

This is the setting I was missing:
$config['db']['schemacheck'] = 'script';
With that added to your config.php, all PEAR errors are instantly gone.

@jpope, I know you gave me your config.php which had this weeks ago. Sorry for my stupidity. I had the answer right there, but for some reason didn't come up with it.

@lnxwalt, @tonybaldwin &co: I can now confirm statusnet 1.1.x pulled today from git (branch 1.1.x, not master!) works like a charm with the latest PHP5.4 from the Arch repos. If any of you was hesitating to switch to 5.4, please do make the upgrade.

Thanks and sorry for the noise!

jpope

Interesting. I can't remember if I had changed that setting when I attempted the 5.4 upgrade or not. I did lighten my instance so I could test things one at a time. May have to give this a go soon. ;)