Topic: SME 7.x (current) - broken at the moment

[cfmunster]

I have been running SME Server for a few months, and I have been happy with it. It has run just fine until this morning.

I tried to help a user connect to her home drive from her desktop this morning, and the system would not accept her login. I tried disconnecting and re-connecting my home drive and found that it was no longer accepting my login, either. I went into the server-manager via HTTPS and updated all the software to the latest revs, and that has made things infinitely worse. After the reboot, Apache refuses to start. I tried accessing the server manager from the console and it says no way.

I suspect that I have installed something (or lots of things) that have created a conflict in the Apache configuration. My first guess would be PHP5, but I'm really not a Linux guru so I could be totally off base. Here is the basic list of stuff I have installed:

- Jinzora
- SugarCRM
- eGroupware
- MySQL myAdmin
- Roundcube

I think along the way I installed PHP5, that's why it jumped out at me.

Any help would be appreciated, even just thoughts on how to get Apache going again. I have access to the shell on the console. I am going to start poking around in the Apache conf and see where that gets me.

Rob

--

UPDATE:

I just checked Apache error_log. The log has a ton of [crit] errors (28) No space left on device: mod_rewrite could not create rewrite_log_lock. I am debugging now, any help is great. Thanks!

[raem]

cfmunster

> I suspect that I have installed something (or lots of things) that have
> I think along the way I installed PHP5, that's why it jumped out at me.

PHP5 will definitely break things when upgrading. Search the forums on "PHP5" for an answer re how to undo the installation and revert to the current PHP

On a sme7.2 server I get:
rpm -qa php*

php-4.3.9-3.22.5
php-ldap-4.3.9-3.22.5
php-domxml-4.3.9-3.22.5
php-pear-4.3.9-3.22.5
php-mysql-4.3.9-3.22.5
php-mbstring-4.3.9-3.22.5
php-gd-4.3.9-3.22.5
php-imap-4.3.9-3.22.5


To investigate your services not running etc.
to show usage
db

to look at curent settings
db configuration show
(which for this database only can be shortened to)
config show
or to be more readable
config show |more
(press space to scroll down)

to look at specific settings

config show httpd-e-smith
   
httpd-e-smith=service
    TCPPort=80
    access=public
    status=enabled

config show httpd-admin

config show squid
squid=service
    EnforceSafePorts=no
    SafePorts=21,70,80,81,119,210,443,563,980,1024-65535
    TCPPort=3128
    TCPProxyPort=80:3128
    TransparentPort=3128
    access=private
    status=enabled

To check if services are running

/etc/init.d/httpd-e-smith status
/etc/init.d/httpd-admin status
/etc/init.d/squid status
/etc/init.d/smb status

& so on for each service listed in the configuration db

To check your account settings

db accounts show username

[cfmunster]

Thanks for all the config/debug details, that helps a lot.
I was wrong about PHP5, I have PHP4 installed. I must have not gone ahead with 5 for some reason.  I am running the same rev. as you.

Checking the HTTP services, e-smith reported down 0 seconds, want up. Admin reported up for 3700+ seconds, normally down.

I figured out I could map drives to the system from my desktop using the admin username and password, so it looks like that is an issue with the LDAP configuration. Could the HTTP service be dying because of the same issue? I changed the workgroup name, I wonder if that change hosed the current user permissions?

Rob

[raem]

cfmunster

You need to look at all your log files in /var/log/...
to find any indications of what's wrong.

eg
/var/log/httpd/error
/var/log/messages
& so on

[raem]

cfmunster

To restart services in sme7.2 do
sv t /service/httpd-e-smith
& so on
look in /service for more service names

man sv for details

Look at the log files at the time of day you try this to see what is reported

[Normando]

cfmunster
See my post, may be help you

http://bugs.contribs.org/show_bug.cgi?id=2035#c38

[guest22]

UPDATE:

I just checked Apache error_log. The log has a ton of [crit] errors (28) No space left on device: mod_rewrite could not create rewrite_log_lock. I am debugging now, any help is great. Thanks!

'No space left on device........' what does 'df -h' on the console tell you....?

guest

[CharlieBrady]

UPDATE:

I just checked Apache error_log. The log has a ton of [crit] errors (28) No space left on device: mod_rewrite could not create rewrite_log_lock. I am debugging now, any help is great. Thanks!

'No space left on device........' what does 'df -h' on the console tell you....?

guest

'df -h' won't tell you anything about IPC shared memory. AIUI, mod_rewrite uses IPC shared memory.

[cfmunster]

I have been away for a week and have not been able to deal with this situation, but I am back now and taking a look at it.

I logged into the console and did:

yum clean all
yum update

That updated all the packages to 7.2, at least that's what is seems to have done. Still no love on the httpd-e-smith service or on users.

Honestly, the biggest problem I have is users. The only users that work for anything any longer seem to be root and admin. I can map to shares on the system using admin, but I can't map to shares using standard user accounts. Maybe this issue is connected to the Apache issue, because e-smith is running as user www, right? I tried to run e-smith as user admin group admin, I have no idea if that should work or not, but it didn't.

Any thoughts? Can someone point me in the right direction for getting the user logins working again?

[mmccarn]

A search for "No space left on device: mod_rewrite could not create rewrite_log_lock" turns up this topic: http://forums.contribs.org/index.php?topic=33171.msg141328#msg141328

[micropitt]

Did you look at the httpd.conf? Is it loading the correct php module? is the php module at the location where httpd.conf is pointing to? Did you make any other changes to the httpd.conf? I had something similar happen on a test server, installed PHP5 and nothing worked after that. I deleted PHP5 and installed PHP4 again with rpm -Uvh --force which then created the correct entrees in the httpd.conf for the php modules to load. If still nothing is working start un-installing contribs. 

[cfmunster]

mmccarn, thanks, I followed that thread's suggestion but that didn't solve my problem. e-smith goes up and down all day long, up for a second, then down, then up for a second, then down. I can tell it is doing this because it write a separate set entries to the error_log each time it cycles.

micropitt, I don't know about the PHP module. Everything in e-smith was working fine until I let the system update the installed RPMs just before I started this thread. After a re-config and reboot, no dice.

The user problem started before I updated the software, I'm guessing when I changed the name of the LDAP directory. Could that change be the cause of my problems in the user system? It seems like a possibility.

Anyway, I am becoming discouraged by the difficulty of troubleshooting the problem and I am considering nuking the whole install and going to Ubuntu. I'm not using the mail server, firewall, directory server (except for access to the machine itself), DNS or DHCP. All I really need is a LAMP stack and Samba shares. My only challenge is migrating a big chunk of files from here to another install.

[cfmunster]

One last try at this. I am downloading the 7.2 CD ISO. I am going to try to get back to a base working state with the CD.

[cfmunster]

One last try at this. I am downloading the 7.2 CD ISO. I am going to try to get back to a base working state with the CD.

Happy happy, joy joy.  I popped in the CD and ran the upgrade from boot. It did almost nothing- the whole process probably took 60 seconds, but when I rebooted, everything was working again. No issues whatsoever.

Sorry to have dragged everyone through the mess, thanks for all the suggestions. And thanks to the SME team for building a solid upgrade installer that fixed my problem!

[cfmunster]

One last note about this issue. The existing users were still hosed- no login, no share access, etc. I only have a handful of accounts on the system and I'm not using it for mail, so I just nuked and recreated the user accounts and home directories and everything works fine again.

[micropitt]

.....and stay away from the LDAP directory. Re-naming the LDAP directory is never a good idea......