spoke too soon (slighly) I notice when playing with roundmail (user login) when I tick on the Account tab I see 'SOAP error cannot connect to host at the bottom of some of the screens'. I see from the troubleshooting FAQ that soap errors seem to be associated with permissions on the remote user but all the permissions are ticked correctly... all the error boxes refer to plugin - ispconfig3_account at the bottom and I notice that the config.inc.php file does not use https to connect. one of the urls is: http://ns9.cdbsystems.com/webmail/?_task=settings&_action=plugin.ispconfig3_account how I added https but does not seem to help. HOWEVER it sort of looks like the screens are working. the password field is shown, information is displayed apparrently correctly. maybe this Soap error is ignorable??
ISPCONFIG3 - resolved but NOT roundcube issues solved LOL ok the change to the main.inc.php in roundcube.conf to remove the '' and put in 'localhost' did NOT really resolve things it allowed logins, but messed up the plugins! answer - put the ---snip from main.inc.php // For example %n = mail.domain.tld, %d = domain.tld $rcmail_config['default_host'] = ''; ---snip default_host BACK to empty ('') - and edit /var/www/apps/roundcube/plugsins/ispconfig3_account/config/config.inc.php to make: $rcmail_config['soap_url'] = 'https://localhost:8080/remote/'; originally it was http NOT https, and had the server name that I had to change to localhost. VOILA - all works. logins succeed and no soap errors and plugins work! might want to ad these as notes to others. 100% fix on roundcube finally.
one last roundcube issue I see the ISPCONFIG3 install installed v0.8.5 and I'm looking at v0.9.2 I'll try the upgrade see how everything works!
roundcube upgrade A-OK running 0.9.2 all worked as indicated so roundcube seems to be happy! now to look at kolab LOL
Perc 6/i Raid controller commands the raid controller is a MegaRaid by LSI. installing the utility (from rpm) it creates the utility in the directory: /opt/MegaRaid/MegaCli a sufficient snapshot command suitable for ISPCONFIG to display under status would be: /opt/MegaRAID/MegaCli/MegaCli64 -AdpAllinfo -aALL this gives All info on All the adapters in the system. how could we put this in ISPCONFIG?
on another issue - owncloud I've installed owncloud (yes, its outside of ISPCONFIG though I did use ISPCONFIG3 to setup the website) - and all is working EXCEPT that there is apparently a bug in php 5.3 that prevents you from uploading more than about a 2gb file. seems it has been fixed in php 5.5 any reason I should fear doing a 5.5 upgrade? ISPCONFIG likely to break? its not available in the normal yum update you have to add a repository rpm -Uvh http://mirror.webtatic.com/yum/el6/latest.rpm then yum install php55w php55w-opcache any known issues with php 5.5 and ISPCONFIG? thanks till
There are a few warnings about use of deprecated functions when you use php 5.5 but it should work as far as I know. Warnings for deprecated functions can be disabled in the php.ini file.
odd behaviour - cgi related?? several of my sites use an obscure cgi-invoked program and I've noticed something recently and I believe this has just started happenening. best demonstrated: http://www.cdbsystems.com/cgi-bin/h...tem/CDBData/Invoices/InvSearch/invsearch.html if you click on this is SHOULD give you a 'page expired' message. just keep refreshing and from time to time you will get a 500 Error. then you refresh it again and you get the correct behaviour. any ideas?
Do you have apache mod_evasive installed? Which error message do you get in the error.log of the website?
error log I get these errors: [Fri Jul 19 13:31:01 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't spawn child process: /var/www/clients/client0/web21/cgi-bin/htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [Fri Jul 19 13:31:39 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't create child process: 12: htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [Fri Jul 19 13:31:39 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't spawn child process: /var/www/clients/client0/web21/cgi-bin/htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [Fri Jul 19 13:31:43 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't create child process: 12: htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [Fri Jul 19 13:31:43 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't spawn child process: /var/www/clients/client0/web21/cgi-bin/htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [Fri Jul 19 13:32:43 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't create child process: 12: htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [Fri Jul 19 13:32:43 2013] [error] [client 108.28.205.205] (12)Cannot allocate memory: couldn't spawn child process: /var/www/clients/client0/web21/cgi-bin/htmlos.cgi, referer: http://www.howtoforge.com/forums/showthread.php?p=299996 [root@ns9 cdbsystems.com]# the webite has CGI checked (as it has to run htmlos.cgi a rather old-school compiled C executable), and php has Mod seleted. the only directives are: <Directory /var/www/clients/client0/web21/cgi-bin> Options Indexes FollowSymLinks MultiViews +ExecCGI Order allow,deny,+Exec Allow from all </Directory> any thoughts? I've got 32gb of ram and dual six-core opterons.
seems to have fixed ... bumped up MaxClients from 256 to 4096 in prefork and worker seems to have made it happy also set ServerLimit to 4096 as well so far unable to reproduce the error. I stayed with Apache2 and not nginx because this is a true compiled executable CGI. any chance nginx might actually run it? might be willing to give it a chance
Have you considered to install munin and monit on the server to monitor and visualize the server usage? Here is a short guide from nginx wiki about running a cgi script with nginx: http://wiki.nginx.org/SimpleCGI But I did not try that, so I dont know if it works. It might also be that apache is not the problem here and the ressources are taken up by the cgi script itself. In any case, if you want to try nginx I would recommend to do this in a virtual machine first as you cant easily switch your apache setup to nginx without reinstalling the server. You can download a perfect server virtual machine which works with virtualbox here from howtoforge.
munit etc alas they cant install due to dependency errors with some man pages (ONLY man pages conflict). I asked if you that it was ok to just --force cdb.
ftp issues with pure-ftpd I'm revisiting problems with ftp uploads - I seem to logoff and login on EVERY uploaded file: --snip-- Aug 3 20:45:56 ns9 pure-ftpd: ([email protected]) [INFO] Logout. Aug 3 20:45:56 ns9 pure-ftpd: ([email protected]) [INFO] New connection from 192.168.2.1 Aug 3 20:45:56 ns9 pure-ftpd: ([email protected]) [INFO] tmm is now logged in Aug 3 20:45:56 ns9 pure-ftpd: ([email protected]) [NOTICE] /var/www/clients/client0/web30//web/index_htm_files/44.png uploaded (1788 bytes, 349.01KB/sec) Aug 3 20:45:59 ns9 named[19571]: client 199.101.162.104#45764: query (cache) 'ieee-consultants.org/MX/IN' denied Aug 3 20:45:59 ns9 named[19571]: client 65.55.37.41#55450: query (cache) 'ieee-consultants.org/A/IN' denied Aug 3 20:46:17 ns9 pure-ftpd: ([email protected]) [INFO] Logout. Aug 3 20:46:17 ns9 pure-ftpd: ([email protected]) [INFO] New connection from 192.168.2.1 Aug 3 20:46:17 ns9 pure-ftpd: ([email protected]) [INFO] tmm is now logged in Aug 3 20:46:17 ns9 pure-ftpd: ([email protected]) [NOTICE] /var/www/clients/client0/web30//web/index_htm_files/[email protected] uploaded (5399 bytes, 494.46KB/sec) Aug 3 20:46:18 ns9 named[19571]: client 199.101.162.105#52241: query (cache) 'ieee-consultants.org/MX/IN' denied Aug 3 20:46:38 ns9 pure-ftpd: ([email protected]) [INFO] Logout. Aug 3 20:46:38 ns9 pure-ftpd: ([email protected]) [INFO] New connection from 192.168.2.1 Aug 3 20:46:38 ns9 pure-ftpd: ([email protected]) [INFO] tmm is now logged in Aug 3 20:46:38 ns9 pure-ftpd: ([email protected]) [NOTICE] /var/www/clients/client0/web30//web/index_htm_files/45.png uploaded (213158 bytes, 2226.34KB/sec) --snip-- any ideas what might be happening?? I've not had this problem with this client (xara) before cdb
Mots likely the logoff is initiated by the client. Test another client e.g. the fireftp plugin for firefox to see if it has the same behaviour.
just for your amusement it turns out the ftp client logoff was due to my software trying to use passive mode - and I had forgotten to put a big block of holes in the firewall (30000-50000) to accomodate that. once the firewall was opened the problem evaporated