IBM DS Series Storage Battery Guidelines
A recent case of a controller battery failure in an IBM DS5020 storage system has been encountered. After the battery failure in one of the controllers, storage performance dropped significantly, seriously affecting normal user business operations. Referencing the "IBM DS4000/DS5000 Quick Maintenance Manual" and the "IBM DS5020 Storage Installation, User, and Maintenance Guide," the causes for performance degradation and steps to replace the battery are as follows:
[Design Principles of DS4000/DS5000 Batteries and Cache]
1. To ensure the safety of write cache data, in the DS4000/DS5000 system, if a battery fails (is in a Failed state) or exceeds the design life specified by the firmware (39 months for firmware below version 06.60, and 10 years for firmware version 06.60 and above), the write cache is automatically disabled. This is a designed mechanism to prioritize data safety.
2. To ensure the safety of write cache data, the Cache Mirroring write cache mirroring function is enabled by default on both controllers. Therefore, when the write cache on one controller is disabled, the write cache on the other controller also becomes unavailable.
[Fault Symptoms After Write Cache Disablement]
1. Application response time at the host slows down. Checking host performance metrics via topas shows a significant increase in I/O waiting time, with each logical disk's busy percentage approaching 100%, yet I/O throughput remains very low, ranging from tens of KB to 1 MB per second:
2. Checking the LUN cache status may also reveal the following situation:
In the Storage Manager client graphical software, the write cache status shows a red dot, indicating that although the write cache is enabled, it is currently not active or is in a Suspended state.
For FAStT200/DS4100 and DS4300 models, since the batteries are installed inside the controllers, replacing the battery requires removing the controller containing the failed battery.
Battery replacement for DS4200/DS4400/DS4500/DS4700/DS5020/DS5100/DS5300 does not require taking the system offline or removing the controller.
For safety reasons, we generally recommend scheduling downtime for offline replacement. If online battery replacement is required, sufficient preparation must be made.
Please use the following procedure to replace the battery unit.
Install the DS Storage Manager client software to view the storage operational status, logs, and alarm information. The default IP addresses for Controller A and B are 192.168.128.101 and 192.168.128.102.
Locate the RAID controller containing the failed battery unit.
Check the indicator lights on the battery units to determine which of the two battery units has failed. Replace only the battery unit indicated as failed.
Put on anti-static protection equipment.
To remove the failed battery unit from the RAID controller, press the orange battery unit sliding latch towards the black battery handle to unlock the battery in the DS5020 chassis, then use the handle to slowly pull the battery unit out of the controller chassis.
Place the removed battery unit on a flat surface.
Open the packaging of the new battery unit. Place the new battery unit on a dry, flat surface. Keep all packaging materials available in case you need to return the new battery unit.
Insert the new battery unit into the controller chassis. Ensure the new battery unit is inserted into the battery unit tray in the correct orientation. Push the battery unit fully into the tray until it clicks into place. Do not force installation. The battery unit is designed to prevent incorrect insertion into the chassis. After the new battery is installed, the green "Battery Charging" indicator light will flash continuously until the battery is fully charged.
After the battery charges from its depleted shipping state to its optimal state, it will perform a self-test cycle to evaluate the battery discharge time. Do not power off the storage subsystem for more than 24 hours to ensure the battery completes charging and finishes the first self-test cycle.
After the new battery is installed, it will take 15 minutes to 72 hours for the new battery to fully charge. Before the battery is fully charged, the controller's write cache will remain automatically disabled.