The following are some of the symptoms of this issue:

• NetScaler SVM is not accessible through GUI or SSH.

• You will not be able to ping SVM IP address.

• The mps_stat.log file shows "No buffer space available"  message. You will see this message during Nitro calls or when you try to ping external IP from SVM.

• The mps_service.log shows "no resource available" message.

• A ping from SVM through console reports "no buffer space available".

To resolve this issue upgrade to the latest NetScaler 10.5 build.

To workaround this issue perform an interface reset from SVM’s FreeBSD console:

1. Log on to Dom0.

2. Use command xl list to display the VM ID.

3. Run command xl console < Management Service VM's ID>.

4. Enter SVM credentials.

5. Enter shell. After entering FreeBSD shell prompt, issue the following command:
   ifconfig <SVM’s management interface number> down
   ifconfig <SVM’s management interface number> up

   Now verify if the SVM is accessible.

6. In case the preceding steps did not resolve the issue then instead of performing a normal reboot of the SVM force a core dump through the XenServer console.
   From the XenServer issue command:
   /usr/lib/xen/bin/crash_guest <dom id of SVM>

Excerpt from NetScaler 10.5 Release Notes:
"On an SDX appliance, the Management Service may lose connectivity. The issue is seen only with Management Service which is in the UP state for many days, minimum being 277 days. - [From Build 58.11] [#444854]"