Unexpected Shutdown every 2 week (at self test I'm sure) Win 2003 - 64bit

[Anonymous user]

Hi,

I receive a pretty nice unexpected shutdown every two week on Sunday. Finally I'm sure that the UPS is the cause because today I figure out that every 2 week after turn on the self test run.

I configure my APC Backup XS BX1500LCD (connect via USB) with the native Windows 2003 Power Management.
(Win 2003 64bit latest updates - used as VMware host)
Don't use APC PowerChute because the free version don't have enough freedom to choose shutdown time. (I need shutdown after 60% battery left)
In the PowerChute I only could configure max after 5min or 6min battery power left or something like this.

Windows Power Management is configured with 76% Low battery (just sound and text) and 61% critical level.(sound, text and shutdown)

My questions is if during self test the APC goes under this level and maybe don't send a right event (because "just" self test) to windows so it could shutdown properly?
Windows Event log don't show any information beside the unexpected shutdown event.
LCD display show Event 1

Please let me know how I can solve my bloody Sunday issue or if you need more information to support my. Thank you very much.

BTW, I know that this UPS hardware is maybe not appropriate for server but we just need a safe controlled shutdown in case of 🙂

This is the closed info i found: Referenced content removed but don't want to reopen old topics.

Message was edited by: Tilo because my English s****

Message was edited by: Tilo

[BillP]

hello

this could be caused by a couple different things..

* do you know how much runtime the UPS is providing? you can find this through the LCD menu? i highly doubt that the UPS is discharging down to 61% during a self test but even so - you should be able to see it on the LCD. if you arent there on a sunday, you can reset the UPS two week counter by power cycling the UPS but then I assume it'd be bad if it shuts down during production.

* How much wattage is the unit providing to your server? this should be on the front LCD display as well since you arent using powerchute.

* can you be 100% sure that this UPS has held up this server during a power outage before. if not, are you able to perform a "pull plug test?" this means that you unplug the UPS from the wall, to simulate a power outage? if so, i'd recommend doing this at a non-critical time because if the server shuts down, it makes me think it doesnt like this UPS for one of a couple reasons - 1.) you want to be 100% sure the server is plugged into a battery back up outlet, 2.) the transfer time to battery from online mode is 4-6ms, some server class machines cannot handle this amount of time and turn off or reboot, or 3.) the server does not like the step approximated sine wave output that the UPS outputs on battery. other higher end UPSs output a pure a sinewave which is what we recommend for servers.
it could also be a combination of the above items that I mentioned. you could also try the pull plug test with a lamp or something similar. if it stays up, you know its something like what i mentioned above.

let us know how you make out! hope this helps.

[Anonymous user]

hi, thank for the fast response.

1. Run time around 85min. And yes Sunday is not to bad for unexpected shutdowns :-). So I will not change this time by turn off and turn on my UPS.
2. Load around 50w, In and out 118v. (hope this is want u need to know)
3. No, I'm not sure my first test was a bit 'unorganized' because had problems with the PowerChute SW. Will test a pull out plug this week during off peak time. With a lamp also connected and switch on during test. Does is make a different if i shutdown the running VMwares so in case of only the host system will crash. Or should the load be similar for this test (guess makes no different to test the reaction time, so if the 4-6ms is fine for this PC-server.
3.1. yes 100% PC is plug in the rear right bottom at the 1500LCD.
3.2 This is actual a normal PC machine (normal motherboard, no redundant Power) so not extra Server HW. Guess the pull plug out test show if the ms are to much for this machine.
3.3 Because its a normal PC this maybe don't matter anymore. If still how can i figure this out.

Thanks for react so fast, I keep you up to date what the pull plug our test show.

[BillP]

1. Run time around 85min. And yes Sunday is not to bad for unexpected shutdowns :-). So I will not change this time by turn off and turn on my UPS.

ok, thats fine.

2. Load around 50w, In and out 118v. (hope this is want u need to know)

yes i am looking to know how much power your computer is drawing. so, ok 50 watts. the unit maxes out at about 860 i believe so we want to make sure its not overloading at any point

3. No, I'm not sure my first test was a bit 'unorganized' because had problems with the PowerChute SW. Will test a pull out plug this week during off peak time. With a lamp also connected and switch on during test. Does is make a different if i shutdown the running VMwares so in case of only the host system will crash. Or should the load be similar for this test (guess makes no different to test the reaction time, so if the 4-6ms is fine for this PC-server.

you can shutdown the VMWares first incase the server crashes. you are right, we need to use the same server for the test FIRST to see what it tries, then try an item like the lamp to see if the essential switching function works.

3.1. yes 100% PC is plug in the rear right bottom at the 1500LCD.

ok, good.

3.2 This is actual a normal PC machine (normal motherboard, no redundant Power) so not extra Server HW. Guess the pull plug out test show if the ms are to much for this machine.

ok, thats fine. maybe it is a high end power supply even though its not redundant. we will see.

3.3 Because its a normal PC this maybe don't matter anymore. If still how can i figure this out.

it may even if its a normal PC. we also need to worry about power factor corrected power supplies but lets not go there yet. let us know when you can do the test 🙂

[Anonymous user]

I run the test tonight and unfortunately figure out that the pull plug out test was not successfully.

1. Test just with a lamp.-> after plug out the lamp was very short a bit flashing.

2. Server without load (beside of the lamp 🙂 )
-> Lamp the same, but server shutdown immediately and try to reboot. But during the reboots the server only reach the Win2003 load up screen not more.
so after the BIOS and RAID load up it somehow go down and stay in the reboot cycle.
Turn off and start (on battery) result the same reboot cycle.
After plug in the UPS power again the server start full and works fine.

The Power Unit in the PC/Server is a
EarthWatts 500 Watt power supply (80 PLUS® certified)
Here all specification. http://www.antec.com/usa/productDetails.php?lan=us&id=27500

Is this a HW or software issue? (because it goes until Win2003 load up screen)

Just read that the Power Unit is maybe the problem:
This is what a guy in a HW review said:

Using with a UPS backup?
Reviewer: ITsjim from Florida on Oct 19, 2008


Consider using something other than the Antec Earthwatts series if you want your UPS to keep it running. I have this 500w PSU that came with an Antec case. Testing soon showed that the UPS shuts down abruptly whenever the UPS switches over. UPS is well sized for the job. Responses from my queries to Antec tell me that the Earthwatts series requires an UPS with pure sinewave -- the more expensive type that are used in server applications. This requirement is not stated anywhere in Antec tech docs. APC (UPSmfr) research has reported that any PSU with active power correction factor (PFC) is likely to shutdown during UPS switchover as small as 8ms. The shutdown is by design and stems from a current inrush during even the smallest switchover time. If you need your box to keep going under UPS power, please consider looking for a passive PFC PSU. Still waiting for Antec decision on exchange.

Any thought about this?

Thanks, and good night 🙂

[BillP · Accepted solution]

hi,

thanks for all the helpful info.

[take a look at this about power factor corrected power supples|http://nam-en.apc.com/cgi-bin/nam_en.cfg/php/enduser/std_adp.php?p_faqid=8883]

if Antec suggested that a UPS with a pure sinewave output be used, you will need to look at a [Smart UPS|http://www.apc.com/products/family/index.cfm?id=165]

i think that you are having the same issue as the other customer unfortunately between the sensitivity on the power supply and the inrush current.