Unified ISSU Concepts

A unified in-service software upgrade (unified ISSU) enables you to upgrade between two different JUNOS software releases with no disruption on the control plane and with minimal disruption of traffic. Unified ISSU is only supported on dual Routing Engine platforms. In addition, the graceful Routing Engine switchover (GRES) and nonstop active routing (NSR) must be enabled.

A unified ISSU provides the following benefits:

• Eliminates network downtime during software image upgrades
• Reduces operating costs, while delivering higher service levels
• Allows fast implementation of new features

Note: The master Routing Engine and backup Routing Engine must be running the same software version before you can perform a unified ISSU. You cannot take any PICs online or offline during a unified ISSU.

To perform a unified ISSU, complete the following steps:

1. Enable Graceful Routing Engine switchover and nonstop active routing. Verify that the Routing Engines and protocols are synchronized.
2. Download the new software package from the Juniper Networks Support Web site and then copy the package to the router.
3. Issue the request system software in-service-upgrade command on the master Routing Engine.

A JUNOS release package comprises three distinct systems:

• Juniper Networks Operating System, which provides system control and all the features and functions of the Juniper Networks router that executes in the Routing Engines
• Juniper Networks Packet Forwarding Engine, which supports the high-performance traffic forwarding and packet handling capabilities
• Interface control

After the request system software in-service-upgrade command is issued, the following process occurs.

Note: In the illustrations, a solid line indicates the high-speed internal link between a Routing Engine and a Packet Forwarding Engine. A dotted line indicates the chassis process (chassisd), another method of communication between a Routing Engine and a Packet Forwarding Engine. RE0m and RE1s indicate master and backup (or standby) Routing Engines, respectively.

Note: The following process pertains to all supported routing platforms except the TX Matrix platform. For information about the unified ISSU process on the TX Matrix platform, see Unified ISSU Process on TX Matrix Platform. On M320 and T320 routers and on T640 and T1600 routing nodes, the Packet Forwarding Engine resides on an FPC. However, on an M120 router, the Forwarding Engine Board (FEB) replaces the functions of a Packet Forwarding Engine. In the illustrations and steps, when considering an M120 router, you can regard the PFE as an FPC. As an additional step on an M120 router, after the FPCs and PICs have been upgraded, the FEBs are upgraded.

1. The master Routing Engine validates the router configuration to ensure that it can be committed using the new software version. Checks are made for unsupported configurations and for unsupported Physical Interface Cards (PICs). Unsupported PICs do not prevent a unified ISSU. The software issues a warning to indicate that these PICs will restart during the upgrade. Similarly, an unsupported protocol configuration does not prevent a unified ISSU. The software issues a warning that packet loss may occur for the protocol during the upgrade.

   

2. When the validation succeeds, the kernel state synchronization daemon (ksyncd) synchronizes the kernel on the backup Routing Engine with the master Routing Engine.
3. The backup Routing Engine is upgraded with the new software image. Before being upgraded, the backup Routing Engine gets the configuration file from the master Routing Engine and validates the configuration to ensure that it can be committed using the new software version. After being upgraded, it is resynchronized with the master Routing Engine. In the illustration, an apostrophe ( ' ) indicates the device is running the new version of software.

   

4. The chassis process (chassisd) on the master Routing Engine prepares other software processes for the unified ISSU. When all the processes are ready, chassisd sends an ISSU_PREPARE message to the Flexible PIC Concentrators (FPCs) installed in the router.
5. The Packet Forwarding Engine on each FPC saves its state and downloads the new software image from the backup Routing Engine. Next, each Packet Forwarding Engine sends an ISSU_READY message to the chassis process (chassisd).

   

6. After receiving an ISSU_READY message from a Packet Forwarding Engine, the chassis process (chassisd) sends an ISSU_REBOOT message to the FPC on which the Packet Forwarding Engine resides. The FPC reboots with the new software image. After the FPC is rebooted, the Packet Forwarding Engine restores the FPC state and a high-speed internal link is established with the backup Routing Engine running the new software. The chassis process (chassisd) is also re-established with the master Routing Engine.
7. After all Packet Forwarding Engines have sent a READY message via the chassis process (chassisd) on the master Routing Engine, other software processes are prepared for a Routing Engine switchover. The system is ready for a switchover at this point.

   

   Note: In the case of an M120 router, the FEBs are upgraded at this point. When all FEBs have been upgraded, the system is ready for a switchover.

8. The Routing Engine switchover occurs and the backup Routing Engine becomes the new master Routing Engine.

   

9. The new backup Routing Engine is now upgraded to the new software image. (This step is skipped if the no-old-master-upgrade option is specified.)

   

10. When the backup Routing Engine has been successfully upgraded, the unified ISSU is complete.

Unified ISSU Process on TX Matrix Platform

After you issue the request system software in-service-upgrade command on a TX Matrix platform, the following process occurs.

1. The management process (mgd) on the master Routing Engine of the TX Matrix platform (global master) checks whether nonstop active routing (NSR) and graceful Routing Engine switchover (GRES) are enabled in the current configuration.
2. After successful validation of nonstop active routing and graceful Routing Engine switchover configuration, the management process copies the new image to the backup Routing Engines on the TX Matrix platform and the T640 routing nodes.
3. The kernel synchronization process (ksyncd) on the backup Routing Engines synchronizes the kernel on the backup Routing Engines with that of the master Routing Engines.
4. The backup Routing Engines are upgraded with the new software and are rebooted. After rebooting, the backup Routing Engines are once again synchronized with the global master Routing Engine.
5. The unified ISSU control moves from the management process to the chassis process (chassisd). The chassis process informs the software processes about the unified ISSU and waits for responses from various software processes (such as spmb).
6. After receiving messages from the software processes indicating that the processes are ready for unified ISSU, the chassis process on the global master Routing Engine sends messages to the chassis process on the routing nodes to start the unified ISSU.
7. The chassis process on the routing nodes sends ISSU_PREPARE messages to the field replaceable units (FRUs), such as FPC and intelligent PICs.
8. On receiving an ISSU_PREPARE message, the Packet Forwarding Engines save the current state information and download the new software image from the backup Routing Engines. Next, each Packet Forwarding Engine sends ISSU_READY messages to the chassis process.
9. On receiving an ISSU_READY message from the Packet Forwarding Engines, the chassis process sends an ISSU_REBOOT message to the FRUs. While the upgrade is in progress, the FRUs keep sending ISSU_IN_PROGRESS messages to the chassis process on the routing nodes. The chassis process on each routing node, in turn, sends an ISSU_IN_PROGRESS message to the chassis process on the global master Routing Engine.
10. After the reboot, the Packet Forwarding Engines restore the saved state information and connect back to the routing nodes; the chassis process on each routing node sends an ISSU_READY message to the chassis process on the global master Routing Engine. The ISSU_READY message from the chassis process on the routing nodes indicates that the unified ISSU is complete on the FRUs.
11. The unified ISSU control moves back to the management process on the global master Routing Engine.
12. The management process initiates Routing Engine switchover on the master Routing Engines.
13. Routing Engine switchover occurs on the TX Matrix platform and the T640 routing nodes.

    Note: Currently, the FRUs on a TX Matrix platform do not support graceful Routing Engine switchover and are rebooted every time graceful Routing Engine switchover occurs.

14. After the switchover, the FRUs connect to the new master Routing Engines, and the chassis manager and PFE manager on the T640 routing node FRUs connect to the new master Routing Engines on the T640 routing nodes.
15. The management process on the global master Routing Engine initiates the upgrade process on the old master Routing Engines on the T640 routing nodes.
16. After the old master Routing Engines on the T640 routing nodes are upgraded, the management process initiates the upgrade of the old global master Routing Engine, that is, the old master Routing Engine on the TX Matrix platform.
17. After a successful unified ISSU, the TX Matrix platform and the T640 routing nodes are rebooted.

Copyright © 2008, Juniper Networks, Inc. All rights reserved. Trademark Notice.