We have had several calls over the past couple of weeks where servers or other equipment have unexpectedly shutdown. Upon investigation we found that a UPS with failing batteries caused the problem. Normally a UPS is responsible for supplying power to critical network devices if there is a power outage. The power can be maintained for from several minutes to an hour or so depending on the capacity of the UPS and the load on it. This is a good thing. We don't want servers crashing when the power goes out.
However, UPSs use batteries and batteries age and fail. We have found that when UPS batteries fail the UPS can do strange things including shutting the power to the connected devices off even if there is no power outage. This is not a good thing!
Because UPSs are quite and not usually highly visible we tend to forget they exist until a problem occurs. To avoid any surprises, UPSs should be tested regularly. The test is simple, make sure no one is connected to the devices that the UPS is connected to, make sure no applications are running on your server. and pull the plug. The UPS is plugged into a power outlet on the wall. Pull the plug from the wall. The devices connected to the UPS should stay running. The UPS will usually start emitting some type of sound indicating power has been lost. After you have confirmed that the UPS can withstand a power outage, plug the plug back into the wall. The beeping should stop. Be careful with this test. You are playing with electricity and it doesn't always play fair. If you are unsure about this test get someone who knows this stuff to do it.
I also suggest that UPSs connected to critical devices be replaced after 5 or more years. They are not that expensive anymore and the value of not having problems caused by a crashed server far out way the investment.
We have included this test in our list of items that we perform for our Maintenance Program Clients.
No comments:
Post a Comment