Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?


Preventing Graceful Routing Engine Switchover in the Case of Slow Disks

Unexpected slow disk access can happen for various reasons—a faulty or bad sector, for example—causing a hold up of the normal operation of processes such as the routing process (rpd). Eventually, the router’s performance will be impacted. Under these circumstances, it may take longer for the typical failover mechanism to be triggered.

Juniper Networks has introduced a disk monitoring daemon to solve this dilemma. The daemon detects slow disk access and initiates failover. Failover can minimize the traffic impact and ease the load on the original primary Routing Engine for its backlog clean up.

However, there are instances when you might not want failover to occur. You might commit a large set of changes or even minor changes that might lead to a series of updates on the routing topology. Such activity could lead to extensive disk access delay and, therefore, trigger failover. For expected disk access delays like this, where you do not want to trigger failover, you can choose to not have failover occur by setting the chassis redundancy failover not-on-disk-underperform configuration command. Another way is to disable the disk monitoring daemon completely by setting the system processes gstatd disable command.

To prevent failovers in the case of slow disks in the Routing Engine:

Set the option for preventing gstatd from initiating failovers in response to slow disks at the [edit chassis redundancy failover] hierarchy level.