A few small APC questions

[0olong]

I have a Smart-UPS connected to a Dell PowerEdge 2600 with a script set-up to shutdown a second server with PowerChute BE 7.04 running. I'd like to test the shutdown sequence, I know that it has a self test, but the bosses want to know if it'll work in an actual situation. Any suggestions on a real-time test?

Also, has anyone had a situation where their getting daily power failure / restores? I seem to be getting one on a daily basis and am trying to see what the causes may be. The failure only lasts two seconds (literally) before power is restored.

 

[fs483]

I usually do a yearly test of my UPS for a complete shutdown. I basically unplug the power from the UPS and let the server run on the UPS alone. I measure the time it takes for the battery to run out before the shutdown sequence begins and I also measure the time it ACTUALLY takes for the server to shutdown. With this data, I can adjust the scripts so that I have enough juice to properly shutdown the server before the UPS runs out. Once the server is shutdown, I plug the power to the UPS back in and see if the server turns back on by itself.

As for daily power failures, I have a few customer who have MANY power failures/spikes/burnouts per day. Some are in a more remote area while some are in town. I get a notification when the power acts up or if I'm in the server room, I usually hear the battery kick in for a few seconds.

 

[newBPhoneGuy]

is it possible that you have the ups to do a self test everyday and would this log as a power failure and restore since it seems to me that for the self test the powerchute would simulate a power failure for a couple seconde the restore its self ill test it on my unit and see but that is my guess for now.....

 

[fs483]

Nope. The self test runs every x number of hours (manufacturer defined). I don't think you can change it. In the power chute management software under device properties, diagnostic and self-test, you'll see when the last self test was done and the status. Some people don't like the APC self test because it totally disconnects itself from the power input and puts the full load on the batteries. If the batteries are defective (or the charging unit), the self test might cause the server to crash or unexpected shutdowns. I have to agree. APC should of designed something allowing it to sustain an equivalent load (under normal working conditions) to be used while it's on it's self test.

On another point. I just had a real world test of one of my upses and it failed miserably. I had a power failure at one of my sites and the APC1500VA tower ups failed within 15 minutes of outage. The last time I tested it was a year ago and it was able to keep the EXACT same load for 45 minutes. The self test didn't discover the problem but the power failure did. After the power came back, the UPS didn't even turn itself back on so someone had to manually power it up. 2 upses have died at this site (probably due to heat in the server room).

I also had a second UPS failure today but this one is not managed by me and don't know which manufacturer. This morning, one of my customer complained they couldn't access the server. After further testing on site, we discovered the UPS used by the building manager in their telecom equipment had failed. We didn't like the idea of having someone else take care of the equipment but we had no choice because all the network cabling in the building went to a shared telecom room for all the tenants. We only had 2 switches in that room but it was enough to cause us 2 hours of downtime.

I wasn't lucky today with UPSes and hopefully will have a better day tomorrow !