Cannot access WUI after trying to backup config

[agentsam]

We were not upgrading firmware during this process. Currently, and have always been "live-testing" on 1.4.2, board revision D.

All that was performed, is that we logged into GUI, and selected Utilities - Backup and the UI hung; we gave it 45 minutes and refreshed, tried a different browser and wouldn't get login screen. SSH remained active. In CLI, reload cold didn't seem to actually power cycle the unit; not sure what it did. Checked config and saw that https was enabled, and correct port; for good measure we re-enabled https port and verified; we then enabled http and port and verified. Lastly, a drop to linux shell and reboot -f was performed and it was found POE 1- 48v was off after power on; took a local tech to visit site and get that turned back on.

As it would appear to only report V, doing the math, current draw looks to be 1A +/- .03A

[sirhc]

OK there is a bug in v1.4.2 that affects some people.

If you have SMTP enabled and there is an communications problems with the SMTP server or your setting as not correct the firmware gets stalled trying to reach the SMTP server and it can cause the loss of the UI, v1.4.3rc4 fixes this.

On the main status page the UI reports the watts being used for each port.
If the watts is 0 and you know it is powering a device that means the current sensors for that port are damaged.
WIth v1.4.2 the switch will assume since there is no current being used that POE is not needed and turns the POE off. This behavior has been changed in v1.4.3rc4 and it will not turn POE off incase your current sensors are damaged.

Current sensors most commonly get damaged from poor grounding. Most common grounding mistake is the equipment box/electric service is not bonded to the tower grounding system. Next most common grounding mistake is people think the steel tower is a good path to ground and do not run dedicated grounding cables.

Here are some good posts on grounding:
viewtopic.php?f=30&t=1816
viewtopic.php?f=30&t=188
viewtopic.php?f=30&t=1429
viewtopic.php?f=17&t=1786&start=30#p13447

[agentsam]

Great, we'll either disable SMTP or ensure communication is established. Explains WUI hang.

As for POE, it is reporting W fine so I don't believe that is the issue as to why port 1 was off, and sensor is still good; any other ideas?

*We do follow all of your guides on grounding, specifically yours (i.e. we send your posts to electricians).

[sirhc]

It is possible I guess that if the SMTP messed up the UI it may have messed up other things? - I will ask Eric to chime in.

I do suggest you try v1.4.3rc4 as it fixes a couple things.

[Eric Stern]

As for POE, it is reporting W fine so I don't believe that is the issue as to why port 1 was off, and sensor is still good; any other ideas?

Doing a "reboot -f" is not a safe reboot, so the switch thought it was a cold boot. During a cold boot if a PoE port is showing a link before power is turned on then PoE is automatically disabled as a safety measure. Since this was not a cold boot the AF on port 1 still had a link, so the switch disabled it. There will be a message about this in the log.

This safety measure was removed in 1.4.3, so it won't happen again.

[agentsam]

Thanks guys.

Can one clarify what "reload cold" actually does; I assume it just reloads current saved config but no power cycle?

Is there a safe way to power cycle via CLI?

As for removing that safety check in 1.4.3, are there extra precautions or checks in place with that removed, or may we find ourselves in more trouble with a reboot -f if that check is not in place?

[Eric Stern]

Can one clarify what "reload cold" actually does; I assume it just reloads current saved config but no power cycle?

It simply reboots the switch. No power cycle.

Is there a safe way to power cycle via CLI?

There is no way to power cycle the switch except to physically disconnect the power.

As for removing that safety check in 1.4.3, are there extra precautions or checks in place with that removed, or may we find ourselves in more trouble with a reboot -f if that check is not in place?

The original intent was to protect against configurations being accidentally applied. ie if you backed up the config from switch A which had PoE enabled on port 1, and restored that config to switch B which had a non-PoE device plugged into port 1. But it caused more problems than it solved so it was removed. It will still log if it occurs, but no longer automatically disables PoE. You should be fine.

[agentsam]

Thanks for the answers; is it by design, or limitation, that one cannot truly power cycle other than physically onsite? Are we perhaps talking different technical applications of terms? Do you consider restarting a PC remotely, via RDP say, power cycling; or if I issue "reload" on a Cisco Catalyst switch, is that power cycling according to your frame of reference? I get confused by "rebooting is not power cycling" as I've always come to use those terms as mutually exclusive.

[sirhc]

Thanks for the answers; is it by design, or limitation, that one cannot truly power cycle other than physically onsite? Are we perhaps talking different technical applications of terms? Do you consider restarting a PC remotely, via RDP say, power cycling; or if I issue "reload" on a Cisco Catalyst switch, is that power cycling according to your frame of reference? I get confused by "rebooting is not power cycling" as I've always come to use those terms as mutually exclusive.

It is the same process as with any device such as a computer.

If you tell your computer to reboot that is not a power cycle.

A power cycle means the power is cut allowing the electronics to fully discharge and reset.

Sometimes a reboot is not enough and all electronics need to discharge and fully reset.

How can a piece of equipment simulate a full power cycle? You would have to have another device in front of it and then it uses relays to cut power then power it back up but then you may need to power cycle the power cycler and this creates a vicious cycle.

ALL A SOFTWARE REBOOT DOES WITH ANY DEVICE IS JUST THAT TELL THE OS TO RELOAD NOT POWER CYCLE

[agentsam]

Thanks for clarifying, I always assumed (wrongly it seems) that there was circuity present within the power supply of a piece of equipment that did exactly that, a relay that did cut the power albeit briefly. In effect, there is thus segregation on the PCB between the power supply circuitry (that does in fact always receive power, as you state) and the rest of the board that does in fact completely loose power during a reboot; you're stating this is definitely not the case? How would one account for the memory clearing characteristics of a reboot, if power isn't being refrained from that part of the circuitry?