I have a homelab with an APC BackUPS 1500 XM, a Synology NAS running DSM 6.2, and a Proxmox server. My desktop Windows 10 PC is plugged into a smaller UPS.<CUT>
If the power goes out, the synology shuts down, and I can either hit Proxmox on the web interface, ssh into it and shut it<CUT>
down, or hit the power button. If it runs down the battery, it goes into sleep mode.
I'm wondering if there's a way to have either the Proxmox system or the Synology system connected to the UPS and shut down
the other one?
I'm wondering if there's a way to have either the Proxmox system or the Synology system connected to the UPS and shut down the other one?
run the UPS usb port into the proxmox host and run N.U.T (networked ups thing). then run a nut client on the other machines.
Mark Hofmann wrote to Kurt Weiske <=-
I'm wondering if there's a way to have either the Proxmox system or the Synology system connected to the UPS and shut down the other one?
I think APC makes software that does that, but likely wants to charge
for it. One UPS power failure can trigger an event to do multiple
things.
I never got that fancy with my setup and I have (10) APC UPS units over here. If my power it out long enough for them to go out, everything
just powers off immediately. I would try and do a clean shutdown, but even if it goes down hard I have not had issues with my ESXi
server/VMs. Worst case, I could restore things.
Mark Hofmann wrote to Charles Blackburn <=-
I typically get around 20 or so minutes of runtime in the event of a
power outage on most of my stuff, which isn't too bad.
I'm wondering if there's a way to have either the Proxmox system or the Synology system connected to the UPS and shutI think APC makes software that does that, but likely wants to charge for it. One UPS power failure can trigger an event to
down the other one?
do multiple things.
run the UPS usb port into the proxmox host and run N.U.T (networked ups thing). then run a nut client on the otherIs there an equivalent for those using APC UPS using the network management cards (IP based)? I monitor all my APC UPS units
machines.
using the IP based management cards.
I typically get around 20 or so minutes of runtime in the event of a power outage on most of my stuff, which isn't too bad.
I never got that fancy with my setup and I have (10) APC UPS units over here. If my power it out long enough for them
to go out, everything
just powers off immediately. I would try and do a clean shutdown, but even if it goes down hard I have not had issues
with my ESXi server/VMs. Worst case, I could restore things.
Hey, if you have 10 UPSes, why not? :)
I'm looking to replace my proxmox laptop with a new SFF desktop I bought, which prompted looking into network UPS support.
The shame is that my laptop CPU only draws 15w, the desktop 65w. I'll get less run time out of the UPS with the new system.
I never got that fancy with my setup and I have (10) APC UPS unitsover MH> here. If my power it out long enough for them to go out, everything MH> just powers off immediately. I would try and do a clean shutdown, but
even if it goes down hard I have not had issues with my ESXi server/VMs. Worst case, I could restore things.
Hey, if you have 10 UPSes, why not? :)
With my current Proxmox + Synology + router setup, I get around 65
minutes of run time -- that's with two printers plugged into the UPS because I just plugged the UPS from the wall into the UPS.
worked awesome here yesterday with this stupid hurricane.... power went out.. ran on ups for 15 minutes and then shut everything down except my cablemodem and firewall (on a separate ups as they dont draw much and can stay up for hours)
worked awesome here yesterday with this stupid hurricane.... power went out.. ran on ups for 15 minutes and then shutIt appears there is a NUT port for Windows and I knew it also works with Home Assistant (which I also use at home).
everything down except my cablemodem and firewall (on a separate ups as they dont draw much and can stay up for hours)
I'll take a look and see how it works. Hopefully there is a web gui and I can see all the UPS units in the house via a
central place.
I do monitor them in Home Assistant using the regular snmp commands to poll the different information.
I found doing the manual shutdown if I'm around worked better, but certainl am not opposed to giving it another shot with different software and option There is likely something better these days that would work with my ESXi se and Windows servers/systems.
i beleive you can use SNMP to pull the mibs, but i havent gotten that far to be honest.. one of those things i picked up
and then put down to do something else lol
Mark Hofmann wrote to Kurt Weiske <=-
With my current Proxmox + Synology + router setup, I get around 65
minutes of run time -- that's with two printers plugged into the UPS because I just plugged the UPS from the wall into the UPS.
Wow, that is decent runtime but the load is way less than my setup.
As I run my own Email server... I take the UPS-emails generated from the APC daughtercard when power goes to battery-state and execute a simple
DOS batchfile which just creates a dummy file in an SMB share.
All of the VM's check every 15 minutes to half-hour for the presence of that dummy file. If it exists, they shutdown. Call it a simple
"heartbeat" event.
Theoretically, I could get even longer run time by taking my Proxmox
server off of the UPS and use NUT to shut it down, since it's a laptop
with a battery. It comes in handy - when I had an extended power outage,
I brought the laptop to work, connected to guest wifi and transferred
mail packets from my uplink to my hosts and vice versa.
I forgot to mention this bit about the home email server. Technically I so of run my own email server, but it is really just a relay these days. I ha it send all mail to my free Mailjet account for delivery.
For ESXi, I believe there are command line actions that would perform shutdowns that could be added to batch files. Mainly because not all my VM are Windows, they are all sorts of stuff that I would want to have ESXi perform a graceful shutdown.
Then the issue becomes the power-on when power is restored. Mainly because would need at least one of my domain controllers to come up first or some other systems would boot up and have issues and need to be rebooted again.
Of course I could just do all this manually like I do today in the event po it out for an extended time. That has not happened here for years, fortunatly.
I've ran my own email system for at least 15... 20 years now I think.
I run Hmailserver as an SMTP front-end for my Exchange server.
The only work I've ever had to do in that timeframe was configure DNSBL correctly in the beginning and later on a few Txt records in DNS for some weird new fad that basically told the world "please don't spam me".
I dunno, I guess since I work intimately with Exchange and Outlook in my career, why not run it at home. Once you disable all the debugging and other crap that comes out of the box its really not that bad, takes up little resources here. I don't like building new VM's every time a new Exchange version comes out but its not like that happens often.
In all this time I've never had a problem delivering email to anyone and have
no incentive to stop as I have a commercial grade fiber Internet connection.
I never really got into the Exsi-specific commands, because as explained once
all the VM's sense that dummy file they shutdown via. the Windows
shutdown command. I'm not running non-Windows stuff here though...
What about the startup-sequence settings in Exsi? List the DC first on
that one server and just use timer sequence on the others?
I have one APC SmartUps 1400 and replace the batteries every couple of years from some local vendor that actually sells them at a decent price. The daughterboards for those APC UPS's are a godsend and I'll never understand why Ethernet or probe-contacts were never made "standard
issue" for their stuff until much later.
I've actually never had a full-cycle on UPS, I'm guessing I'll get about
a half hour or so of runtime with all the stuff in the rack.
Non-essential stuff just goes to the main PDU... I don't need NAS storage during an outage.
Sysop: | deepend |
---|---|
Location: | Calgary, Alberta |
Users: | 255 |
Nodes: | 10 (0 / 10) |
Uptime: | 153:19:39 |
Calls: | 1,724 |
Calls today: | 4 |
Files: | 4,107 |
D/L today: |
10 files (9,986K bytes) |
Messages: | 392,941 |