Edge SWG devices running an SGOS 7.3 version prior to 7.3.14.2 that have been up for 49 days or longer may experience throughput or connection issues. Access log delivery utilizing continuous mode with persistent TCP connections may find logging slow down or have connections to the logging server reset. In addition, customers may also experience TCP disconnects or slow performance with other applications under the same circumstances.
The underlying network stack in SGOS relies upon a timing variable counter of 1ms. This counter is primarily used to make timing calculations for TCP/IP related to round trip time and idle flow timeouts.
In SGOS version 7.3.x, these timing variables were permitted to be stored as a 64-bit data type, but in some cases the counters were compared against 32-bit values that were reset after 49.7 days.
This issue can surface in several ways:
SGOS 6.7.x versions are not affected.
This issue affects all versions of SGOS 7.3.x up to and including version 7.3.14.2.
The issue only occurs after the Edge SWG device has been running for 49 days.
The issue can cause idle TCP persistent connections to be reset or a reduction of throughput for some TCP connections.
In most cases, end users will not notice any issues but there may be exceptions depending on the application.
Broadcom has corrected this issue in the following versions
7.3.14.x where x is equal to or greater than 3.
7.3.x where x is equal to or greater than 15.
7.4.x where x is equal to or greater than 2.
To prevent this issue, reboot Edge SWG devices before a 49-day uptime.