UPS Time is incorrect

nilesf wrote:

Running ReadyNAS Pro 4 with firmware 4.2.19. Last night around 00:07:10 (am) I lost power and at 00:07:24 (am) power returned. The e-mail sent to me is not stating the correct time and it show that I lost power for 6 seconds which is also incorrect. I have the NTP set to my Active Directory domain controller's IP address which has been on time since its incarnation. UPS is on battery power. [Fri Mar 9 00:07:27 EST 2012] UPS is on line power. [Fri Mar 9 00:07:33 EST 2012] I have two UPS, one for the ReadyNAS Pro 4 and one for my AD server. My AD server is running PowerChute Business Edition and the times of 00:07:11 (am) (on battery) and 00:07:24 (am) (on AC) are correct. This a 14 second time on battery compared to 6 seconds that the ReadyNAS is reporting. 8 seconds differance is not a big deal but it just goes to show that that the ReadyNAS is off by 8 seconds with its reporting. Please advise, Niles

Why do you care about this?

I don't, but for those who are time sensative this maybe an issue for them. It also draws attention to what if this was a longer duration of "on-battery"? How will the ReadyNAS respond if a 14 second duration is multiplied into hours? Will the ReadyNAS multiply the offset differance? Just a ten fold would mean an entire minute.

Assuming a 10 fold:
PowerChute would be 14*10=140/60=2.3xxx (round to 2.3 minutes)

ReadyNAS would be 6*10=60 or 1 minute.

If this is the case then this could be a very big problem for those who rely on the ReadyNAS to report correctly. It also shows the lack of quality in the code.


Niles

nilesf wrote:

I don't, but for those who are time sensative this maybe an issue for them. It also draws attention to what if this was a longer duration of "on-battery"? How will the ReadyNAS respond if a 14 second duration is multiplied into hours? Will the ReadyNAS multiply the offset differance? Just a ten fold would mean an entire minute. Assuming a 10 fold: PowerChute would be 14*10=140/60=2.3xxx (round to 2.3 minutes) ReadyNAS would be 6*10=60 or 1 minute. If this is the case then this could be a very big problem for those who rely on the ReadyNAS to report correctly. It also shows the lack of quality in the code. Niles

Very speculative. You are for some reason assuming PowerChute is correct in its reporting and the ReadyNAS is at fault. Do you have evidence that supports that conclusion?

You have no evidence at all to support your theory that the error is somehow proportional to the outage time. The NAS is getting this information from the UPS, so the two UPS's are the devices that signal the power up/power down event. It is quite possible that the NAS (and maybe PowerChute) are polling for status, and that the error is simply a result of the polling interval. This seems like a more plausible explanation to me, it is at least equally plausible.

If you want to gather proper evidence, you have a very easy way to proceed. Just unplug the UPS for a while, then plug it back in. Track the time, and see what both devices report. It shouldn't be difficult to measure what the error is, and how it relates to outage duration. (Best to have a backup before you deliberately create power issues).

In any case, the signal that really matters is the battery life remaining, since that is what drives the clean shutdown (which preserves the data in the RAID array). So perhaps as part of your testing you should let the power completely drain from the UPS and confirm that the NAS in fact does the proper shutdown. It would be prudent to ensure that no disk writes are being done during that test. So maybe disconnect the NAS from the network, and make sure no backup jobs are running.

Agree that this is speculative to some degree. I also agree that PowerChute can be at fault. You need to cover all possible scenarios.

I should have given more detail to support my post, therefore I do apologize. I have several servers that have their own UPS running PowerChute. Server hardware and UPS are all identical and all have very basic roles. They all reported the same time to the second of the power failure. Now if even if one was off by a second then my assumption would have no support.

Now if we were to throw in the recent coronal mass ejection from the sun then all of this would be relevant. Five years ago I was in a server room that was actually shut from such an ejection. So anything is possible.

nilesf wrote:
Now if we were to throw in the recent coronal mass ejection from the sun then all of this would be relevant. Five years ago I was in a server room that was actually shut from such an ejection. So anything is possible.

Wow. Must have made for an interesting status report!

Anyway, for your own piece of mind I suggest you might want to do some more UPS testing with the NAS, just to confirm that it is working ok.

StephenB wrote:
nilesf wrote: Now if we were to throw in the recent coronal mass ejection from the sun then all of this would be erelevant. Five years ago I was in a server room that was actually shut from such an ejection. So anything is possible.

Wow. Must have made for an interesting status report!

You have no idea. Fortunately the UPSs had sufficent backbone to support their infrastucture. I will test jsut to satisfy this post. Pleae give me some time to complete a few test later this weekend.