Article ID: 130222, created on Jan 27, 2017, last review on Jan 30, 2017

  • Applies to:
  • Operations Automation

Symptoms

Different symptoms for the same issue are observed based on VE OS:

  • Debian-8.0:

    Task for Provisioning "Panel_6_1" for APS application Plesk in VPS Panel fails with the error:

    APS Application Error:
    500 Internal Server Error
    [Pvps::App::Panel::Exception::RuntimeException] Internal Server Error: Plesk autoinstaller has been launched but the process has finished with unexpected error. For more details see autoinstaller log '(/var/log/plesk/install//autoinstaller3.log)'.Please fix the problems and re-schedule the task or try to start autoinstaller manually.
    

    /tmp/autoinstaller3.log inside the provisioned VE contains errors:

    invoke-rc.d: initscript spamassassin, action "start" failed.
    dpkg: error processing package spamassassin (--configure):
     subprocess installed post-installation script returned error exit status 1
    
  • CentOS-7.0:

    Panel provisioning task fails with:

    APS Application Error:
    500 Internal Server Error
    [Pvps::App::Panel::Exception::RuntimeException] Internal Server Error: Unable to initialize Plesk. The running of command ('/usr/local/psa/bin/init_conf' --init ...
    

    /tmp/autoinstaller3.log inside the provisioned VE contains errors:

    Configuring Postfix Plesk Mail Driver...
     Trying to comment out comsat service record from /etc/xinetd.conf...  there is no comsat service record in /etc/xinetd.conf
    Fatal error: plesk::mail::postfix::PostfixConfigurationError(postmap: fatal: parameter inet_interfaces: no local interface found for ::1
    

There is a big amount of OACI virtual environments with IPv6 addresses assigned to them. The corresponding harware node with the faulty VE hosted on it contains the following errors in dmesg in large amount:

[2781621.011489] Neighbour table overflow.

Cause

::1 address is not configured inside the VE on the loopback adapter, due to insufficient amount of IPv6 neigh cache storage on the hardware node.

[root@vz ~]# grep . /proc/sys/net/ipv6/neigh/default/gc_thresh*
/proc/sys/net/ipv6/neigh/default/gc_thresh1:128
/proc/sys/net/ipv6/neigh/default/gc_thresh2:2048
/proc/sys/net/ipv6/neigh/default/gc_thresh3:4096

Resolution

For a permanent solution, increase the size of IPv6 neigh cache:

# echo "# Tuning IPv6 neigh cache" >>  /etc/sysctl.conf
# echo "net.ipv6.neigh.default.gc_thresh1 = 4096" >> /etc/sysctl.conf
# echo "net.ipv6.neigh.default.gc_thresh2 = 8192" >> /etc/sysctl.conf
# echo "net.ipv6.neigh.default.gc_thresh3 = 16384" >> /etc/sysctl.conf
# sysctl -p

This change will prevent the problem from appearing again.

To fix the failed task, restart the VE and resubmit the task.

In case task resubmission does not help, cancel the sales order and generate a new one. Due to the unstable nature of the issue, another attempt is likely to get provisioned with no issues.

5356b422f65bdad1c3e9edca5d74a1ae caea8340e2d186a540518d08602aa065 e12cea1d47a3125d335d68e6d4e15e07

Email subscription for changes to this article
Save as PDF