Workarounds

[IgorG]

plesk9-kav eztemplate for SuSE-11.0-x86_64 fails installation

Plesk 9.2.2 SuSE 11.0 92090714.19 openSUSE 11.0 (X86-64)

Error like:

[root@pvclin64 ~]# vzpkg install 792157 plesk9-kav
Loading "rpm2vzrpm" plugin
Setting up Install Process
Setting up repositories
Reading repository metadata in from local files
Parsing package install arguments
Resolving Dependencies
--> Populating transaction set with selected packages. Please wait.
---> Downloading header for psa-kav to pack into transaction set.
psa-kav-1.0.0-suse11.0.bu 100% |=========================| 233 kB 00:01
---> Package psa-kav.i386 0:1.0.0-suse11.0.build92090714.19 set to be updated
--> Running transaction check
--> Processing Dependency: libgcc_s.so.1(GLIBC_2.0) for package: psa-kav
--> Processing Dependency: libgcc_s.so.1(GCC_3.3) for package: psa-kav
--> Processing Dependency: libgcc_s.so.1 for package: psa-kav
--> Processing Dependency: libgcc_s.so.1(GCC_3.0) for package: psa-kav
--> Restarting Dependency Resolution with new changes.
--> Populating transaction set with selected packages. Please wait.
---> Downloading header for libgcc43 to pack into transaction set.
libgcc43-4.3.1_20080507-6 100% |=========================| 9.2 kB 00:00
---> Package libgcc43.i586 0:4.3.1_20080507-6.1 set to be updated
--> Running transaction check
...
--> Processing Dependency: libgcc_s.so.1(GCC_3.0)(64bit) for package: libstdc++43
--> Processing Dependency: libgcc_s.so.1()(64bit) for package: boost
--> Processing Dependency: libgcc_s.so.1(GCC_3.0)(64bit) for package: psa
--> Finished Dependency Resolution
...
Error: Missing Dependency: libgcc_s.so.1(GCC_3.0)(64bit) is needed by package yast2-core
Error: Missing Dependency: libgcc_s.so.1(GCC_3.0)(64bit) is needed by package libstdc++43
Error: Missing Dependency: libgcc_s.so.1()(64bit) is needed by package boost
Error: Missing Dependency: libgcc_s.so.1(GCC_3.0)(64bit) is needed by package psa
Error: /usr/share/vzyum/bin/yum failed, exitcode=1

Solution:

http://kb.odin.com/en/7058

 

[IgorG]

Can't change Personal FTP Repository Settings

Plesk 9.2.3

Trying to apply new ftp backup login settings in Plesk Panel produced the following errors in log:

[root@www ~]# tail -f -n0 /var/log/sw-cp-server/error_log
ftp_filemng: error: cURL execution failed: couldn't connect to host

System error 9: Bad file descriptor

As result there is unable to save ftp settings and create backups on ftp server.

Solution:

http://kb.odin.com/en/7059

 

[IgorG]

Empty "from" field and from=#@[]

Plesk 9.2.3

If the email could not be delivered (domain does not resolves or anything else) the mail is returned back to sender with empty "from" field, it could not be delivered by this reason (User_unknown) and qmail sent it to postmaster@localhost with filed "from=#@[]"

Solution:

http://kb.odin.com/en/8097

 

[IgorG]

backup-unpack SyntaxError: invalid syntax

Plesk 9.x

Unable to run backup-unpack:

r# /opt/psa/bin/backup-unpack
File "/opt/psa/bin/backup-unpack", line 5
import backup-unpack
^
SyntaxError: invalid syntax

Solution:

http://kb.odin.com/en/7066

 

[IgorG]

Event handler is not triggered when update domain hosting settings

Event handler does not work for changing 'Use a single directory for housing SSL and non-SSL content' option in domain hosting settings.

How to reproduce:

1)Add event handler as follows:

Domain hosting settings updated
highset (100)
root
/sbin/updated

2) Put the following content into /sbin/updated (or any other)

#!/bin/bash

echo -e "Triggered event handler at `date +%F.%H%M%S`" >> /var/log/messages

3) Check\uncheck 'Use a single directory for housing SSL and non-SSL content'

Actual result: Not triggered

Solution:

http://kb.odin.com/en/8098

 

[IgorG]

[autoinstall.plesk.com]Need to put correct packages for Centos-5-i386

Plesk 8.6.0 (seems 9.2.3 affected too)

autoinstall.plesk.com offers packages for FC6 in http://autoinstall.plesk.com/PSA_8.6.0/update-rpm-CentOS-5-i386/, whcih confilicts with system packages on several CentOS versions - 5.2, 5.3 and 5.4

Solution:

remove bind-libs-9.3.6-4.P1.el5
remove bind-utils-9.3.6-4.P1.el5
install db4-utils.i386 0:4.3.29-10.el5
install gcc-java.i386 0:4.1.2-46.el5 (needed for tomcat component)

 

[IgorG]

Problem with web application management

Plesk 9.2.2

How to reproduce:

a.

1. Create a domain in PPP.
2. Install any application to Document Root (/) directory. i.e. phpBB v3.0.4
3. Notice - it is not possible to add web applications at all on this domain anymore.
4. Create a subdomain for this domain.
5. Now the option to "install web application" is available for the domain, it is possible to install a web application in ANY subfolder.

b.

1. Create a domain and subdomain for it.
2. Notice that it is possible to choose whether to install web application to domain or subdomain.
3. Install any web application to Document Root (/) directory on the domain.
4. Now it is not possible to install web application on subdomain (option 2. is missing)

Solution:

http://kb.odin.com/en/8081

 

[IgorG]

reconfigrator.pl does not overwrite zzz_horde/atmail_vhost.conf

Plesk 9.2.3

After changing server IP address run reconfigurator.pl to make Plesk configured on a new IP. It is all OK but webmail remains linked to OLD IP.

How to reproduce:

1) create a Plesk container with assigned IP;

-bash-3.2# less /etc/httpd/conf.d/zzz_atmail_vhost.conf | grep -i Virtual
<VirtualHost 10.55.62.8:80>
</VirtualHost>
<VirtualHost 10.55.62.8:443>
</VirtualHost>

2) remove this IP and add a new one:

[root@pvclin64 ~]# vzctl set 921 --ipdel all --ipadd 10.55.62.13 --save
Deleting IP address(es): 10.55.62.8
Running command: /etc/sysconfig/vz-scripts/vz-net_del
Adding IP address(es): 10.55.62.13
Running command: /etc/sysconfig/vz-scripts/vz-net_add
Run script /etc/sysconfig/vz-scripts/dists/scripts//redhat-add_ip.sh
Saved parameters for Container 921
[root@pvclin64 ~]# vzlist 921
CTID NPROC STATUS IP_ADDR HOSTNAME
921 40 running 10.55.62.13 plesk922.plesk.ru

3) run reconfigurator.pl to update Plesk configuration:

[root@pvclin64 ~]# vzctl enter 921
entered into Container 921
-bash-3.2# /usr/local/psa/bin/reconfigurator.pl ip_map.txt
Attempting to connect to MySQL: ok
IP map file template ip_map.txt was successfully created
Edit it manually to reflect your configuration
-bash-3.2# vim ip_map.txt
2# /usr/local/psa/bin/reconfigurator.pl ip_map.txt
Using mapfile ip_map.txt - start checking
Checking done, now Plesk configuration will be updated.
10.55.62.8 -> venet0:10.55.62.13 255.255.255.255
Update ip 10.55.62.13 in database... ok
Reconfiguring DNS settings:
domain cvnvc.xcvb.xcvbxc... ok
Restarting DNS service... ok
Now reconfigure apache... ok
Mapping procedure successfully done

4) update vhost configuration:

-bash-3.2# /usr/local/psa/admin/bin/websrvmng -a -v

5) see that webmail vhosts are still pointed to old IP:

-bash-3.2# less /etc/httpd/conf.d/zzz_atmail_vhost.conf | grep -i Virtual
<VirtualHost 10.55.62.8:80>
</VirtualHost>
<VirtualHost 10.55.62.8:443>
</VirtualHost>

As result reconfigurator.pl should update VirtualHost value for zzz_horde/atmail_vhost.conf too.

Solution:

http://kb.odin.com/en/8007

 

[IgorG]

Add "Allow IP change" option to Plesk Linux

Support Plesk clients with dynamic IPs is available in Windows versions only (Allow IP change checkbox in Sessions Preferences) Login to Linux Plesk GUI from ISP who switches IPs every minute (dynamic IPs). Once IP changes, your session to Plesk panel is dropped. You have to re-login to Plesk after every several clicks, which make work in Plesk impossible. Dynamic IPs support should be done in Linux Plesk versions as it is done in Windows Plesk.

Solution:

Unfortunately, there is no ability to change this behavior in config​.
In attach prepared for Plesk 8.6 and 9.x a custom fix, to install it:

1. Copy attached class.Session.php to /usr/local/psa/admin/plib/
2. Execute: mysql -uadmin -p`cat /etc/psa/.psa.shadow` -Dpsa -e "insert into misc (param, val) values ('disable_check_session_ip', 'true');"

Please, don't forget to backup the original file.

'disable_check_session_ip' parameter in psa.misc table is used to enable or disable client's IP check:
- 'true' value means 'do not check client's IP during session initialization'
- other values or parameter absence means 'check client's IP during session initialization'

 

[IgorG]

Fileserver module cannot add/remove user to existing share

Plesk 9.2.3

Samba module produces error when trying to add/remove user from existing share.
Steps to reproduce:

1) enable 'display_errors' and 'log_errors' in /usr/local/psa/admin/conf/php.ini and set 'error_reporting' to 'E_ALL' and restart 'sw-cp-server'

2) Go to PPP on Samba fileserver module page and create two users on 'Users' tab

3) Go to 'Shares' tab and create new share; in 'Access' section choose 'Selected only' radio button and add one user; press OK

4) Edit just created locale and try to add another user to the list of allowed users

As result Error message will be displayed:

====
Catchable fatal error: Object of class FileServerUser could not be converted to string in /usr/local/psa/admin/plib/modules/fileserver/FileServerShare.php on line 234
====

and no update to config of the share is done.


Solution:

http://kb.odin.com/en/8088

 

[IgorG]

maillog file is growing and not processed correctly by DailyMaintenace/script.php

Plesk 9.2.3

maillog file is growing and not processed correctly by DailyMaintenace/script.php

# /usr/local/psa/admin/sbin/statistics --calculate-one --domain-name=jkfasgfsadk.hgs
statistics: idn_toascii() failed: Invalid length
statistics: idn_toascii() failed: Invalid length

statistics failed on calling idn_toascii() with 'Invalid length' error.

Error is caused by the following line from maillog:

Jul 30 15:25:23 94 imapd: 1248963923.114020 LOGOUT, ip=[127.0.0.1], rcvd=17, sent=98, maildir=/etc/httpd/conf.d/ssl.conf

Base maildir is '/var/qmail/mailnames/'. 'statistics' skips length of this base maildir in '/etc/httpd/conf.d/ssl.conf' and treats '.conf' as domain name. But '.' isn't permitted in domain names as first char and idn_toascii() fails on this.

Solution:

Attached fixed 'statistics' for CentOS 5 x32.

 

[IgorG]

enabled global handlers break system aliases in qmail

Plesk 9.2.3

PROBLEM DESCRIPTION: delivery to system aliases (root, postmaster etc.) triggers bounces if global handler such as DomainKeys sign/check is enabled. Messages that generated for mailname to which system alias dereferences to receives empty mails.

ADDITIONAL INFORMATION:

When there is no domain whose name is server's hostname, then any attempt to send mail to system alias (e.g. root) will finish in error since user 'alias' will be used to invoke 'qmail-local' and this user has no access to handler's hooks. Those errors cause bounces and queued messages contain only 'Received' and 'Delivered-To' headers.

In case if domain with name equal to server's hostname exists, then 'qmail-local' is executed with user 'popuser' who has access to handler hooks, however system aliases does not work since domain is virtualdomain -- if there is no 'root' mailbox/alias on domain created, message will be bounced back with 'This_address_no_longer_accepts_mail' error.

STEPS TO REPRODUCE:

0) Set up global handler for checking or signing mail with DomainKeys

1) Add hostname to /var/qmail/control/locals if it is not there and restart Qmail

2) Send mail to 'root' -- 'echo test | mail -s test root' -- and gather logs

3) Add domain whose name is `hostname -f`

4) Send mail to 'root' again and gather logs

5) Remove domain and disable global handler

6) Send mail to root again and gather logs. You will see that now no bounces are generated and everything works fine.

ACTUAL RESULT:

here is full log gathered on step #2:

message sent to mailname that '[email protected]' points to:

[root@ibutorin ~]# cat /var/qmail/queue/mess/1/104825123
Received: (qmail 32430 invoked by alias); 12 Nov 2009 12:56:56 +0900
Delivered-To: [email protected]

EXPECTED RESULT:

Delivery to system aliases should work.

WORKAROUND:

add 'alias' user to 'popuser' group (important to make it main group of user 'alias'):

usermod -g popuser alias

It may be risky for security, but at least in this case it works.

BETTER WORKAROUND:

root@www:/opt/psa/handlers/hooks# chmod o+rx *
root@www:/opt/psa/handlers/hooks# ls -l
total 440
-r-xr-xr-x 1 root popuser 32304 2009-10-09 13:52 autoresponder
-r-xr-xr-x 1 root popuser 28408 2009-10-09 13:52 deliver-quota
-r-xr-xr-x 1 root popuser 35960 2009-10-09 13:55 dk_check
-r-xr-xr-x 1 root popuser 35984 2009-10-09 13:55 dk_sign
-r-sr-xr-x 1 drweb popuser 181240 2009-10-09 13:57 drweb
-r-sr-xr-x 1 root popuser 48864 2009-10-09 13:55 grey
-r-xr-xr-x 1 root popuser 32768 2009-10-09 13:53 spam
-r-xr-xr-x 1 root popuser 25568 2009-10-09 13:55 spf
-r-xr-xr-x 1 root popuser 12648 2009-10-09 13:52 test

 

[IgorG]

Error produced by mchk on newer versions on CentOS

Plesk 9.2.x

'mchk' produces errors due to syntax error in file /usr/local/psa/admin/bin/courier_authpsa_configure .

[root@netsoft-2 courier-imap]# /usr/local/psa/admin/sbin/mchk --with-spam
==> Checking for: mail_spam_restore... ok
==> Checking for: mailsrv_conf_init... ok
==> Checking for: mail_mailbox_restore... ok
==> Checking for: mailsrv_entities_dump... ok
==> Checking for: mail_admin_aliases... ok
==> Checking for: mail_auth_dump... ok
==> Checking for: mailman_lists_dump... ok
==> Checking for: mail_responder_restore... ok
==> Checking for: mail_drweb_restore... not exsists
==> Checking for: mail_kav_restore... not exsists
==> Checking for: mail_spf_restore... ok
==> Checking for: mail_dk_restore... ok
==> Checking for: mail_grey_restore... ok
awk: cmd. line:50: (END OF FILE)
awk: cmd. line:50: invalid char '?' in expression
unable to process "pop3d"
awk: cmd. line:50: (END OF FILE)
awk: cmd. line:50: invalid char '?' in expression
unable to process "pop3d-ssl"
awk: cmd. line:50: (END OF FILE)
awk: cmd. line:50: invalid char '?' in expression
unable to process "imapd"
awk: cmd. line:50: (END OF FILE)
awk: cmd. line:50: invalid char '?' in expression
unable to process "imapd-ssl"

Patch:

--- /usr/local/psa//admin/bin/courier_authpsa_configure.orig 2009-11-14 17:02:36.000000000 +0100
+++ /usr/local/psa//admin/bin/courier_authpsa_configure 2009-11-14 16:57:46.000000000 +0100
@@ -77,8 +77,7 @@
changed=110;
}
exit changed;
-}\
-' $conf >>$temp
+}' $conf >>$temp
}

remove_authpsa()

 

[IgorG]

EmuWebmail cannot find required perl modules

Plesk 8.6.0

Installed application EmuWebMail 7.0.1-7 onto domain, when you open the main application page, it fails with error

Can't locate Net/SMTP_auth.pm in @INC (@INC contains: /home/httpd/vhosts/bhsclass81/lib/i386-linux-thread-multi /home/httpd/vhosts/bhsclass81/lib /usr/lib/perl5/5.8.5/i386-linux-thread-multi /usr/lib/perl5/5.8.5 /usr/lib/perl5/site_perl/5.8.5/i386-linux-thread-multi /usr/lib/perl5/site_perl/5.8.5 /usr/lib/perl5/site_perl /usr/lib/perl5/vendor_perl/5.8.5/i386-linux-thread-multi /usr/lib/perl5/vendor_perl/5.8.5 /usr/lib/perl5/vendor_perl .) at emumail.fcgi line 191.
BEGIN failed--compilation aborted at emumail.fcgi line 191.

Solution:

​The reason of the bug is Perl compiled with wrong @INC
Net:SSLeay perl module should be installed manually from CPAN.

 

[IgorG]

Statistics fails with segmentation fault if first line in access_log empty

Plesk 9.2.3

If there's an empty line in access_log or access_log.processed, statistics fails with segmentation fault.

Solution:

http://kb.odin.com/en/8392

 

[IgorG]

Webmail stops working after configuring wildcard subdomain

Plesk 9.2.3

Webmail stops working after configuring wildcard subdomain

Steps to reproduce:

- Enabled Webmail for the domain and make sure it works
http://webmail.check.cf

- Added a subdomain zz-wildcard.check.cf which should serve as a catch-all (wildcard) subdomain.

# cat /var/www/vhosts/check.cf/subdomains/zz-wildcard/conf/vhost.conf
ServerAlias *.check.cf

# /usr/local/psa/admin/sbin/websrvmng -a -v

Checked http://webmail.check.cf - now it refers to Plesk Default Page despite on the fact webmail configs supposed to be loaded after all vhosts configs:

# grep Include /etc/httpd/conf/httpd.conf
Include conf.d/*.conf

# ls -la /etc/httpd/conf.d/
....
-rw-r--r-- 1 root root 3232 Nov 25 14:05 zz010_psa_httpd.conf
-rw-r--r-- 1 root root 2497 Nov 25 13:57 zzz_horde_vhost.conf

# grep Include /etc/httpd/conf.d/zzz_horde_vhost.conf
Include /etc/psa/webmail/horde/conf.d/*

# cat /etc/psa/webmail/horde/conf.d/check.cf
ServerAlias webmail.check.cf
#

Solution:

Need to load an apache webmails configs before main vhosts confg(zz010_psa_httpd.conf).
Just replace attached utility for centos5x86 and switch-on/switch-off any installed webmail. Or rename zzz_*.conf webmail configs to zz001_*.conf

 

[IgorG]

If /tmp is full error 'FTP network error' is shown when attempting to copy backup

Plesk 9.2.2 Centos 5x86

When selecting backup on 'Personal FTP repository' and clicking 'Copy to ServerRepository' button, temporary file is written in /tmp directory. Usually, size of /tmp partition is not big (or it can be even tmpfs mounted) and once /tmp is full Parallels Plesk Control Panel produces error in interface:

====
Error: FTP network error
====

Temporary file should be stored in DUMP_D/tmp since there are plenty of space.

# grep DUMP_D /etc/psa/psa.conf
DUMP_D /var/lib/psa/dumps

Solution:

Fixed pmm-ras for Centos 5x86 in attach. You can to mount unionfs filesystem on /tmp directory as workaround.

 

[IgorG]

All certificates are reset when domain is restored

Plesk 9.x

Certificates from domains' repositories set on exclusive IP addresses are re-set when any domain's backup is restored.

STEPS TO REPRODUCE:

1) create domain on exclusive IP, create self-signed certificate in this domain's certificate pool and assign this certificate in domain's web hosting settings.

2) create second domain -- no matter whether it is on exclusive or shared IP or whether it uses its own certificate (in this case second domain uses its own certificate on exclusive IP, however the issue has been reproduced with domain
on shared IP with default certificate):

mysql> SELECT d.name, ip.ip_address, c.name FROM domains d, hosting h, IP_Addresses ip, certificates c WHERE c.id=ip.ssl_certificate_id AND ip.id=h.ip_address_id AND h.dom_id=d.id AND d.name LIKE 'cbt%';
+----------+------------+------+
| name | ip_address | name |
+----------+------------+------+
| cbt1.tld | 10.55.62.1 | cbt1 |
| cbt2.tld | 10.55.62.7 | cbt2 |
+----------+------------+------+
2 rows in set (0.00 sec)

3) create backup of second domain. then restore it

ACTUAL RESULT

all domains on exclusive IPs that had certificates from domain's certificate
pool assigned are re-set to 'default certificate' including just restored
domain:

mysql> SELECT d.name, ip.ip_address, c.name FROM domains d, hosting h, IP_Addresses ip, certificates c WHERE c.id=ip.ssl_certificate_id AND ip.id=h.ip_address_id AND h.dom_id=d.id AND d.name LIKE 'cbt%';
+----------+------------+---------------------+
| name | ip_address | name |
+----------+------------+---------------------+
| cbt1.tld | 10.55.62.1 | default certificate |
| cbt2.tld | 10.55.62.7 | default certificate |
+----------+------------+---------------------+
2 rows in set (0.00 sec)

In restore.xml:

---
<migration-dump content-included="true" base="C:\Program Files
(x86)\Parallels\Plesk\Backup\domains/dom2.w2k3-91.plesk.ru/qwe_dom2.w2k3-91.plesk.ru_info_0905131937.xml"
dump-version="9.0" agent-name="Plesk">
<admin id="1" guid="EC6B97C2-8532-4A95-A58B-223D7415F150"
deployer-action="auxiliary">
...
</admin>
<server>
<properties />
</server>
</migration-dump>
---

In migration.log:

---
[19:47:12|INFO: 1004.log] Setting of certificates for IP addresses
[19:47:12|INFO: 1004.log] Request is ready for: LANG=en_US.UTF-8
ALLOW_WEAK_PASSWORDS=1 C:\Program Files (x86)\Parallels\Plesk\bin\ipmanage.exe
'--reset_ssl_certificate'
[19:47:12|INFO: 1004.log] Request string:
ARG[0]=ipmanage&ARG[1]=--reset_ssl_certificate&ENV[ALLOW_WEAK_PASSWORDS]=1&ENV[LANG]=en_US.UTF-8&STDIN=
[19:47:13|INFO: 1004.log] HTTP status code is: 200
[19:47:13|INFO: 1004.log] HTTP response: <cli>
<ok/>
<stdout>Default SSL certificate for all IP addresses was set</stdout>
<return>0</return>
</cli>

SOLUTION:

Backup and replace /usr/local/psa/admin/sbin/deployer

 

[IgorG]

All certificates are reset when domain is restored. Part 2

Additional attachments.

 

[IgorG]

psa-courier-imap package modifies PATH and breaks YaST on SuSE 11

PROBLEM DESCRIPTION:

psa-courier-imap-3.0.8-suse11.0.build90081117.17.i386
psa-courier-imap-3.0.8-suse11.0.build90081117.17.x86_64

After installing psa-courier-imap-3.0.8-suse11.0.build90081117.17 the following error occurs:

The file /etc/profile.d/courier-imap.sh modified the $PATH variable.

# echo $PATH
//sbin://bin:/sbin:/usr/sbin:/usr/local/sbin:/root/bin:/usr/local/bin:/usr/bin:/bin:/usr/X11R6/bin:/usr/games:/usr/lib/mit/bin:/usr/lib/mit/sbin

/etc/profile.d/courier-imap.sh add //sbin://bin to the $PATH variable. The double-slash in the path variable have a negative impact to yast:

# yast
//sbin/yast: line 27: //lib/YaST2/bin/yast2-funcs: No such file or directory
//sbin/yast: line 238: set_lang_from_sysconfig: command not found
//sbin/yast: line 267: check_ncurses: command not found
package yast2-qt is not installed
package yast2-gtk is not installed
Something is wrong with the YaST user interface.

SOLUTION:

http://kb.odin.com/en/6588