MTA unable to start due to wrong /dev/null permissions: Difference between revisions

(Created page with " =MTA unable to start due to wrong /dev/null permissions= {{KB|{{ZC}}|{{ZCS 8.6}}|{{ZCS 8.5}}|{{ZCS 8.0}}|}} <!-- Wiki/Article Body --> ===Problem=== There are two symptom...")
 
No edit summary
 
(23 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 
{{BC|Community Sandbox}}
__FORCETOC__
<div class="col-md-12 ibox-content">
=MTA unable to start due to wrong /dev/null permissions=
=MTA unable to start due to wrong /dev/null permissions=
{{KB|{{ZC}}|{{ZCS 8.6}}|{{ZCS 8.5}}|{{ZCS 8.0}}|}}
{{KB|{{ZC}}|{{ZCS 8.6}}|{{ZCS 8.5}}|{{ZCS 8.0}}|}}


=Problem=
MTA unable to start due to wrong /dev/null permissions. There are two symptoms that may indicate that:


<!-- Wiki/Article Body -->
* 1. Switching to a zimbra user results in the following output:
 
===Problem===
 
There are two symptoms that may indicate that:
 
1. Switch to a zimbra user results in the following output:
 
  -bash: /dev/null: Permission denied
  -bash: /dev/null: Permission denied
  -bash: /dev/null: Permission denied
  -bash: /dev/null: Permission denied
Line 37: Line 34:
  failed.
  failed.


2. The second symptom is that MTA will not start.  
* 2. The second symptom is that MTA will not start.  
 
=Cause=
The problem might be caused by different things. One could be that the permissions are changed during boot time by a security software. Another cause might be that udev is missing a rule, or the rule granting rights to the /dev/null file are changed due to an upgrade of udev to systemd-tools.
 
=Resolution=
To fix the permissions, we can do the following:


* 1) As root, delete the currently existing /dev/null:
# rm /dev/null</code>


===Cause===
* 2) Recreate it with the '''mknod''' command:
# mknod -m 0666 /dev/null c 1 3


The problem might be caused by different things. One could be that the permissions are changed during boot time by a security software. Another cause might be that udev is missing a rule, or the rule granting rights to the /dev/null file is changed due to an upgrade of udev to systed-tools.
=Troubleshooting=
'''1.''' '''W'''e can check in /lib/udev/rules.d the 50-udev-default.rules file (CentOS release 6.7), and make sure there is a line like the following or similar if its a different distribution:
KERNEL=="null|zero|full|random|urandom", MODE="0666"


===Troubleshooting===
'''2.''' '''I'''f that line is there, but the rights are still wrong, we can create persistent audit watch rule, in order to determine whether anything is changing /dev/null permissions after boot.


Potentially we can check what changed
* a) Set up a persistent audit watch in '''/etc/audit/audit.rules''' which will capture any write or attribute changes that are made to /dev/null. Include a '''key''' such as '''null-watch''' in order to easily search later for audit messages related to this rule. Please note, this rule needs to be included at some point after the '''-D''' rule at the top of that file which removes rules initially before proceeding systematically through the file.
===Resolution===
Ideally our rule should be after the line: ''# Feel free to add below this line. See auditctl man page''.


  The first one is more obvious
add the following line:
  -w /dev/null -p wa -k null-watch


'''w''' - write to the file
'''a''' - change in the file’s attribute       


  set in a udev rule
* b) Once this is in place, restart the auditd service in order to query the rules file and initiate this watch:
  service auditd restart


$ grep null /lib/udev/rules.d/*
* c) Check that the rule is in place:
/lib/udev/rules.d/50-udev-default.rules:KERNEL=="null|zero|full|random|urandom", MODE="0666"
# auditctl -l
LIST_RULES: exit,always watch=/dev/null perm=wa key=null-watch


KERNEL=="null|zero|full|random|urandom", MODE="0666"
* d) Then check that auditing is enabled:
------------------
# auditctl -s
Since UDEV isn't correctly changing the permissions for the current /dev/null, it might be possible to try re-creating /dev/null using the mknod command.
AUDIT_STATUS: enabled=1 flag=1 pid=1826 rate_limit=0 backlog_limit=320 lost=0 backlog=0


As root, delete the currently existing /dev/null:
* e) If not, enable auditing and ensure that the auditd service is configured to run persistently, in the appropriate run level:
auditctl -e 1
chkconfig --list auditd


Raw
* f) Reboot the system
# rm /dev/null</code>
Recreate it with the mknod command:


Raw
* g) After the reboot, if the /dev/null has wrong permissions, check for any audit messages related to the watch using the following command:
# mknod -m 0666 /dev/null c 1 3
ls -al /dev/null  
The /dev/null file should now have with the correct permissions.
ausearch -i -k null-watch


If the problem persists after a reboot, then follow the steps in the 'Diagnostic Steps' section of this article, in order to set up an audit watch which may help determine whether /dev/null permissions are being changed after boot.
----
{{Article Footer|Zimbra Collaboration 8.6, 8.5|04/22/2016}}
{{NeedSME|Teodor|SME2|Copy Editor}}

Latest revision as of 16:20, 26 April 2016

MTA unable to start due to wrong /dev/null permissions

   KB 22604        Last updated on 2016-04-26  




0.00
(0 votes)

Problem

MTA unable to start due to wrong /dev/null permissions. There are two symptoms that may indicate that:

  • 1. Switching to a zimbra user results in the following output:
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
-bash: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 138: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 143: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 148: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 153: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 158: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 163: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 168: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 173: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 178: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 220: dpkg: command not found
/opt/zimbra/libexec/get_plat_tag.sh: line 226: /dev/null: Permission denied
/opt/zimbra/libexec/get_plat_tag.sh: line 263: /dev/null: Permission denied 
Can't open perl script "/dev/null": Permission denied
-bash: /dev/null: Permission denied
/opt/zimbra/bin/zmswatchctl: line 45: /dev/null: Permission denied
Starting swatch.../opt/zimbra/bin/zmswatchctl: line 69: cannot redirect standard input from /dev/null: Permission denied
/opt/zimbra/bin/zmswatchctl: line 45: /dev/null: Permission denied
failed.
  • 2. The second symptom is that MTA will not start.

Cause

The problem might be caused by different things. One could be that the permissions are changed during boot time by a security software. Another cause might be that udev is missing a rule, or the rule granting rights to the /dev/null file are changed due to an upgrade of udev to systemd-tools.

Resolution

To fix the permissions, we can do the following:

  • 1) As root, delete the currently existing /dev/null:
# rm /dev/null
  • 2) Recreate it with the mknod command:
# mknod -m 0666 /dev/null c 1 3

Troubleshooting

1. We can check in /lib/udev/rules.d the 50-udev-default.rules file (CentOS release 6.7), and make sure there is a line like the following or similar if its a different distribution:

KERNEL=="null|zero|full|random|urandom", MODE="0666"

2. If that line is there, but the rights are still wrong, we can create persistent audit watch rule, in order to determine whether anything is changing /dev/null permissions after boot.

  • a) Set up a persistent audit watch in /etc/audit/audit.rules which will capture any write or attribute changes that are made to /dev/null. Include a key such as null-watch in order to easily search later for audit messages related to this rule. Please note, this rule needs to be included at some point after the -D rule at the top of that file which removes rules initially before proceeding systematically through the file.

Ideally our rule should be after the line: # Feel free to add below this line. See auditctl man page.

add the following line:

-w /dev/null -p wa -k null-watch

w - write to the file a - change in the file’s attribute

  • b) Once this is in place, restart the auditd service in order to query the rules file and initiate this watch:
service auditd restart
  • c) Check that the rule is in place:
# auditctl -l
LIST_RULES: exit,always watch=/dev/null perm=wa key=null-watch
  • d) Then check that auditing is enabled:
# auditctl -s
AUDIT_STATUS: enabled=1 flag=1 pid=1826 rate_limit=0 backlog_limit=320 lost=0 backlog=0
  • e) If not, enable auditing and ensure that the auditd service is configured to run persistently, in the appropriate run level:
auditctl -e 1
chkconfig --list auditd
  • f) Reboot the system
  • g) After the reboot, if the /dev/null has wrong permissions, check for any audit messages related to the watch using the following command:
ls -al /dev/null 
ausearch -i -k null-watch

Verified Against: Zimbra Collaboration 8.6, 8.5 Date Created: 04/22/2016
Article ID: https://wiki.zimbra.com/index.php?title=MTA_unable_to_start_due_to_wrong_/dev/null_permissions Date Modified: 2016-04-26



Try Zimbra

Try Zimbra Collaboration with a 60-day free trial.
Get it now »

Want to get involved?

You can contribute in the Community, Wiki, Code, or development of Zimlets.
Find out more. »

Looking for a Video?

Visit our YouTube channel to get the latest webinars, technology news, product overviews, and so much more.
Go to the YouTube channel »


Wiki/KB reviewed by Teodor SME2 Copy Editor Last edit by Jorge de la Cruz
Jump to: navigation, search