Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials

Looking for a SolarWinds SNMPv3 Compatible Temperature Monitor/Probe Recommendation

I’m having issues with SolarWinds and AVTECH’s Room Alert 12S Environmental Monitor when using SNMPv3. Firmware on 12S is upgraded allowing SNMPv3 to be utilized. When running SMNPv3, SolarWinds and the temp monitor don’t sync. They will sync with ICMP and lower versions on SNMP. After multiple calls with AVTECH and SolarWinds, it has been determined that the issue is the AES Mode AVTECH uses. AVTECH uses CFB and our SolarWinds representative said our version of SolarWinds will not support CFB for security reasons. Unless AVTECH pushes out new firmware changing the AES Mode to something SolarWinds likes, SNMPv3 with the AVTECH monitors and SolarWinds won't work. As it stand now. I may have to explore using other temperature probes/environmental monitors.

Is anyone else on the forum currently using a different type of temperature/environmental monitor with SolarWinds utilizing SMNPv3? Is so, how well does the probe/system work and would you recommend the product?

Find more posts tagged with

Accepted answers

All comments

keoki

I have solved the temperature question many many times. In one lab, I plot every temperature probe I can find in all of my Cisco switches, and plot them on trending charts, and sometimes apply some calibration factors to them. I HAVE used Avtech units, and I didn't really like them.

Another possible strategy is to monitor the intake temperature of each computer, you should have several intake temps on most servers, including one on each power supply. I think I was getting as many as 25 or 30 temperatures from each server, some in the cpu cores, some in the drive bays, others in other places. With a properly configured SNMP stack you should be able to monitor the intake temperatures of your server power supplies at minimum, and use that to make a measurement of every "U" of used rack space as a measure of cooling system health. Each hard drive has temperature sensors you can probe with SMART commands as well. These temperatures will not be calibrated, but you can come up with methods of either applying corrections, or just plotting the trend and use visual clues to tell good from bad, then set thresholds accordingly.

My favorite solution is to create a sensor network using Dallas Semiconductor's DS18B20 precision temperature probe. I can buy the chips at $1 each on Amazon, and I can get them encapsulated in waterproof pigtails for ~$2. You can put up to 4000 on a single pair of twisted pair, so up can link all of your sensors on a cat5/6 cable string up to 1000ft long, and you can have as many strings of sensors as you need. Each sensor has a serial number, and you can associate those with the location they live. Once you have your string(s) of sensors, you plug them into an iButton 1wire to USB adapter, and plug that into a linux system. For a hither price than the iButton usb adapter, you can get the iButton ethernet adapter... I like the ethernet adapter, but after needing to add this to multiple datacenters, I opted for the cheaper usb part. In the linux system you run the owfs package. Owfs is "one wire file system". Once you have run owfs, all of your temperature probe serial numbers appear in a directory on your linux system that you designate. In that directory each probe appears as a folder with the serial number as the name, Inside that folder are files, each file named for the data it contains. So reading a temperature from your probes is as simple as looking at a named file in a named folder. Once I have this data I send it to my monitoring system for thresholding and alerting. The times I have built these I did them using an open source system, not SolarWinds, but it would be the same thing under SolarWinds. To get the data into SolarWinds you could create an SNMP OID array on your host with that data for SNMP polling OR you could output the readings as SNMP traps or even syslog events that you then ingest into SolarWinds with the appropriate method.

Once I had the data in my telemetry system, I would create a map of the room/building and drop the temperature readings in a green/yellow/red color as appropriate for the temperature thresholds you assign each color to.

You can buy strings of temperature probes wired together that are designed to be incorporated into computer racks, and you just chain them together across the top of your racks.

I have been building datacenters for 35+ years, and you cannot have a single temperature probe and expect it to be meaningful. In a datacenter you have spots that will be 120°f and spots that will be 55°f. If you don't have hot/cold containment, those hot and cold spots can move around unpredictably. The best way to confirm cooling in a datacenter is to probe your sensitive racks from top to bottom on the cold air intake side. Then you can average them together for a "whole room" reading, and you can alert on them individually to help troubleshoot hot spots. Some gear takes cold air in from the front and blows the hot out the back, others pull it in on the right and exhaust it to the left, where you might have another device using that as the intake... so when you rack up gear it is part of your job to know where it gets it's cooling from and not place that in the output of hot exhaust. Hot/cold aisles, hot aisle containment, cold aisle containment, and various other room layout strategies can be used to insure efficient heat rejection from the room. But no matter how your do it, you need to accept that there is no one temperature for your datacenter, and instead embrace the fact that each computer has a hot and a cold side, and temperature needs to be measured there, preferably on the cold side, but monitor the hot side in at least a couple of places too.

For the Much less than the price of one flakey Avtech unit, you can monitor your datacenter temperatures from edge to edge,

Fred.Loucks

Keoki, thank you for the response.

When it comes to temperature monitoring equipment, I am stuck using a preapproved equipment list for monitoring devices. AVTech is on that list. Most likely, I am not going to be able to get away with using semiconductors.

I'll explore some of the other options you presented. Our facilities have hot and cold aisles and heat clouds are nominal based on air flow. Using and plotting information from equipment intakes would work. I'd have to figure out how to size it down though since I have a thousand or so devices I need to deal with around world wide

Again, thanks for the response.

keoki

Yeah, when I was monitoring via the servers, I think I was using the IPMI SNMP support, so it only applied to my newer servers, and older servers I had to manage from the OS side, which meant each server was a separate challenge to get working. My new servers all got the IPMI configured for my MAAS server management system, and I just added a step to add the snmp config to that process using an expect script that I hacked into the onboarding process.

There are thermometers all over the place, you just have to get lucky and have them shared with the snmp daemon.

My chip solution is pretty low level, but as homegrown as it seems, the fact you can buy rack harnesses pre-wired means a lot of people are thinking the same thing. It also doesn't meet your snmpv3 requirements in most cases. Avtech did have 3 probes for each of their units, so you could spread it out a little, and they also had water sensors you could run through all of your condensate pans or other places water might see water leaks caused by your air chillers.

But I forgot to mention the yolink modules I am using now.. Now yolink is not an snmp solution, and to get real-time out of their system you need to write ifttt automations... the downside is it is a cloud application, but it works well for me despite that. It has a variety of module types, and seems stable. I have used it with temperature water and motion sensors. If you don't use ifttt, it only updates hourly except with an alert, so you pretty much have to ifttt is you want to use it for critical tthings. Again, does not meet the snmpv3 requirement.