resets without loss of power

[christopher_mad]

Summit 670 resets when we have a power outage. The device has two power supplies. Both of which are plugged in to separate UPS's. The logging on the UPS shows that that voltage levels never dropped below 118vac and the load on the UPS was constant. We have a number of other devices on the ups. All of which do not restart when the UPS switches to battery support. On battery Support we have about 1.5 hours of battery backup. Even if the power outage is less than one minute before our generator restores power to the minute the Switch still reboots.

10/10/2015 10:58:48.42 System is stable. Change to warm reset mode
10/10/2015 10:58:47.35 Watchdog enabled
10/10/2015 10:58:46.30 snmpMaster process has be en restarted.
10/10/2015 10:58:46.28 Node State[3] = OPERATIONAL
10/10/2015 10:58:44.80 Internal PSU-2 is disconnected.
10/10/2015 10:58:44.80 Internal PSU-1 is disconnected.
10/10/2015 10:58:42.85 snmpMaster initialization com plete
10/10/2015 10:58:42.78 Node State[2] = STANDBY
10/10/2015 10:58:42.78 Node INIT DONE ....
10/10/2015 10:58:42.28 Node State[1] = INIT
10/10/2015 10:58:42.07 DOS protect application started succe ssfully
10/10/2015 10:58:42.01 **** tftpd started *****
10/10/2015 10:58:41.86 snmpSubagent initialization complete
10/10/2015 10:58:41.78 Hal initialization done.
10/10/2015 10:58:41.74 **** telnetd started *****
10/10/2015 10:58:41.56 Internal PSU-2 is powered off.
10/10/2015 10:58:41.56 Internal PSU-2 is present.
10/10/2015 10:58:41.56 Internal PSU-1 is powered off.
10/10/2015 10:58:41.56 Internal PSU-1 is present.
10/10/2015 10:58:41.38 Network Login framework has been initializ ed
10/10/2015 10:58:41.36 Module in fan slot 3 is inserted
10/10/2015 10:58:41.07 Module in fan slot 2 is inserted
10/10/2015 10:58:40.80 telnetd listening on port 23

10/10/2015 10:58:40.79 Module in fan slot 1 is inserted
10/10/2015 10:58:40.49 Starting hal initialization ....
10/10/2015 10:58:40.23 DM started
10/10/2015 10:58:40.22 The Node Manager (NM) has started proc essing.
10/10/2015 10:58:39.90 EPM Started
10/10/2015 10:58:39.89 Booting after System Failur e.
10/10/2015 10:58:38.76 Changing to watchdog warm reset mode

PowerSupply 1 information: State : Powered On
PartInfo : Internal PSU-1 1417W-80303 800462-00-05
Output 1 : 12.04 V, 1.65 A (12V/550W Max)
Power Usage : 30.56 W

PowerSupply 2 information:
State : Powered On
PartInfo : Internal PSU-2 1428W-80690 800462-00-05
Output 1 : 12.08 V, 3.75 A (12V/550W Max)
Power Usage : 69.69 W

since the logs are not very specific. We Swapped out one power supply and then the other. We then swapped out the switch completely. This is still occuring. with all new hardware. Is there a setting that could be causing these reboots?

[EtherMAN]

one trick we use on unmanaged ups or suspicious power problems in cases like this is have one psu on house power and one psu on ups then unplug the ups... If pdu that is on ups resets then you have a faulty or undersized ups.... You can repeat on other ups ... You can do this without an outage due to having one psu still up on house power... This is also a work around if you have a none networked ups or a ups you don't manage in knowing if there is a power event as protected PSU will stay up and unprotected one will go down.. Good luck on your quest for answers ...

[JAMES_WIEDEL]

Hi Christopher,
It looks like you are indeed loosing power to the switch even though it is connected to a UPS.
Since you didn't provide much information about the UPS, this is conjecture on my part:
I would surmise that the UPS during normal operation directly connects the inputs to the outputs. When a power failure occurs, a mechanical switch (relay or contactor depending on who you talk to)changes state to connect the output to the inverter which is driven by the batteries. I have seen some UPSs get slower doing that transfer as batteries age. When new, there is no issue but when some years old, the transfer time is slow enough the switch's power supplies can't hold the load up long enough.
Note that the power supply hold time (not the UPS's) depends on what is going on with the switch at the time of power outage. It there a lot of traffic? Are there lots of VoIP phones, especially if people are talking at the time? POE cameras, especially Pan Tilt Zoom (PTZ)? All of these add to the power drain on the switch's power supplies and the UPS.
You can get a definitive answer to this if you can get a small UPS to connect to ONE of the switch's power supplies. (If those are 1400W power supplies, you will need a 2200VA or so UPS.)
When the next power event occurs, if the switch doesn't go down, look at the logs and see if the power supply without the added UPS goes down. If so you have your answer.
If that power supply does not go down, then you can scratch you head some more!
I suspect that the other items connected to the UPS have longer hold over times so they can ride through short (50ms) transfer times without crashing.
James