Issues with traffic passing through Security Gateway with CoreXL Dynamic Dispatcher enabled

Support Center > Search Results > SecureKnowledge Details

Solution ID	sk108432
Product	CoreXL
Version	R77.30
OS	Gaia, SecurePlatform 2.6, IPSO 6.2, Crossbeam XOS
Platform / Model	All
Date Created	29-Oct-2015
Last Modified	09-Apr-2017

Symptoms

Issues with traffic passing through Security Gateway with CoreXL Dynamic Dispatcher enabled (per sk105261).
Output of "top" command during the issue shows that each CoreXL FW instance (fw_worker_<N>) is consuming only a slight percentage of CPU usage.
"kernel: BUG: soft lockup - CPU#0 stuck for 10s!" appears repeatedly in /var/log/messages file.

In addition, the soft lockup print the stack related to CoreXL FW instances (fw_worker_<N>):
```
Pid: ..., comm:          fw_worker_<N>:/:
... ... ...
EIP is at fwmultik_pending_conn_list_handle_seq+...
```
After reboot, issue occurs again after several weeks of uptime.

Cause

CPU cores that are allocated to CoreXL SND are highly utilized due to iterating over an internal list that holds sequence numbers of the packets that were processed by CoreXL (this internal list is used to support a temporary connection system, required for connection stickiness). The longer a machine is up and running, the longer this internal list becomes, which in turn requires more CPU resources.

Solution

Note: To view this solution you need to Sign In .