HP ProLiant DL288 ISS Technology Update, Volume 7 Number 1 - Newsletter - Page 11

Models with redundant fans, Failure behavior of models with redundant fans - proliant ml350

Page 11 highlights

ISS Technology Update Volume 7, Number 1 Models with redundant fans Several ProLiant DL and ML server models support fan sets that provide cooling redundancy. Redundant fans enable these models to operate indefinitely in the event of a single fan failure. If a fan fails, the necessary logging and alerting actions still occur and the other fans will increase in speed to maintain appropriate cooling for the entire system. The server models that support redundant fans are listed in Table 5-1. Table 5-1. ProLiant DL and ML server models that support redundant fans ProLiant DL servers ProLiant ML servers DL360 DL365 ML350 DL380 DL385 ML370 DL580 DL585 ML570 NOTE: The ML350 G1 through G5 servers have fan configuration offerings stated as redundant; however, these have a non-redundant fansink for the CPU with redundant (non-processor) cooling. Failure behavior of models with redundant fans ProLiant DL and ML servers with an operational health driver implement graceful shutdowns 60 seconds after it is determined that appropriate cooling cannot be maintained. Graceful shutdowns are cancelled if the condition that triggered them is corrected prior to the countdown completion. Beginning with G5 Intel processor-based systems and G2 AMD processor-based systems, graceful shutdowns can occur without a health driver installed. The shutdown is handled by iLO in such a case. If Insight Agents are installed and configured appropriately, they will send a message that describes the event and whether action was taken to shutdown the system. Table 5-2 identifies several failure scenarios and the resulting system behavior. Table 5-2. Failure behavior of ProLiant DL and ML servers with redundant fans Scenario Behavior* Single fan fails during POST The error is logged in the Integrated Management Log (IML1). The health LED turns to amber. The system continues to run OK. Multiple fans fail during POST The error is logged in the IML. The system boots to the point that the ROM checks the fan configuration. ROM will terminate POST and restart the server. Single fan fails; OS with health driver (thermal shutdown enabled) The error is logged in the IML by iLO. The error is logged to the OS logs by the health driver if it is installed. The health LED turns to amber. The system continues to run. Multiple fans fail after POST The error is logged in the IML by iLO. The error is logged to the OS logs by the health driver if it is installed The health driver initiates a graceful shutdown. Single/Multiple fan(s) fail (thermal shutdown disabled) The error is logged in the IML by iLO. The error is logged to the OS logs by the health driver if it is installed. The system continues to run until thermal sensors reach the hardware protection limit, at 1 The Integrated Management Log (IML) is a feature of the ASM controller which keeps a rotating, first-in-first-out record of events in nonvolatile memory. The log can be accessed while the server is operational or through the Integrated Lights-Out user interface. 11

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16

ISS Technology Update
Volume 7, Number 1
11
Models with redundant fans
Several ProLiant DL and ML server models support fan sets that provide cooling redundancy. Redundant fans enable these
models to operate indefinitely in the event of a single fan failure. If a fan fails, the necessary logging and alerting actions still
occur and the other fans will increase in speed to maintain appropriate cooling for the entire system. The server models that
support redundant fans are listed in Table 5-1.
Table 5-1.
ProLiant DL and ML server models that support redundant fans
ProLiant DL servers
ProLiant ML servers
DL360 DL365
ML350
DL380 DL385
ML370
DL580 DL585
ML570
NOTE:
The ML350 G1 through G5 servers have fan configuration offerings stated
as redundant; however, these have a non-redundant fansink for the CPU with
redundant (non-processor) cooling.
Failure behavior of models with redundant fans
ProLiant DL and ML servers with an operational health driver implement graceful shutdowns 60 seconds after it is determined
that appropriate cooling cannot be maintained. Graceful shutdowns are cancelled if the condition that triggered them is
corrected prior to the countdown completion. Beginning with G5 Intel processor-based systems and G2 AMD processor-based
systems, graceful shutdowns can occur without a health driver installed. The shutdown is handled by iLO in such a case. If
Insight Agents are installed and configured appropriately, they will send a message that describes the event and whether action
was taken to shutdown the system. Table 5-2 identifies several failure scenarios and the resulting system behavior.
Table 5-2.
Failure behavior of ProLiant DL and ML servers with redundant fans
Scenario
Behavior*
Single fan fails during POST
The error is logged in the Integrated Management Log (IML
1
).
The health LED turns to amber.
The system continues to run OK.
Multiple fans fail during POST
The error is logged in the IML.
The system boots to the point that the ROM checks the fan configuration.
ROM will terminate POST and restart the server.
Single fan fails; OS with health driver
(thermal shutdown enabled)
The error is logged in the IML by iLO.
The error is logged to the OS logs by the health driver if it is installed.
The health LED turns to amber.
The system continues to run.
Multiple fans fail after POST
The error is logged in the IML by iLO.
The error is logged to the OS logs by the health driver if it is installed
The health driver initiates a graceful shutdown.
Single/Multiple fan(s) fail
(thermal shutdown disabled)
The error is logged in the IML by iLO.
The error is logged to the OS logs by the health driver if it is installed.
The system continues to run until thermal sensors reach the hardware protection limit, at
1
The Integrated Management Log (IML) is a feature of the ASM controller which keeps a rotating, first-in-first-out record of events in non-
volatile memory. The log can be accessed while the server is operational or through the Integrated Lights-Out user interface.