Initial Insights Were Actionable
In a previous post, I mentioned how I was unhappy with the temperatures I was seeing in my home server, so I ordered some new hardware for it. That arrived within a few days, and so I've already tried some different operating systems and NAS servers as a permanent replacement.
I'm still going through pros and cons of each, so I haven't made a decision regarding the operating system to choose on the new server, but there were a couple of components I bought that aren't reliant on the new hardware, so I spend a few minutes last night performing incremental upgrades.
The new components to upgrade that server are a couple of Noctua NF-A12x15 PWM fans. I thought I'd replace the old case fan, and add a fan specifically to cool the hot-swap drive cage.
The old fan was replaced easily, with the Noctua fan using its supplied rubber mountings rather than screws. Then onto the drive cage fan - there is no mounting in my case for a fan to blow air directly across the drives. However, there are some ventilation cut-outs in the drive cage, which happen to be in roughly the right places for fan mounting, so I was able to position the other fan using the rubber mountings.
The motherboard in that NAS box is an ASRock A330ION, which is a 10-year-old design for low-power small-form-factor enthusiast server. It is equipped with 4 fan headers.The motherboard is shipped with a chipset fan, but the CPU is passively cooled.
.jpg)
The case fan was connected to "CHA_FAN1", the chassis fan connector at the top-right of this photo. The CPU fan connector (to the right of the chipset fan connector, between the CPU heatsink and the RAM slots) and the chassis fan connector "CHA_FAN2" (at the bottom of the picture) were not connected. I replaced the existing chassis fan CHA_FAN1 and I attached my HDD fan to CHA_FAN2. Although the fans are PWM, the fan headers are all 3-pin, so the fan speeds will only be controlled by BIOS fan curve settings.
I started the server, and a big puff of dust was expelled! Two things about this: 1) it must be more effective then before, or it wouldn't have expelled the dust. 2) Eww perhaps I should have disassembled and rebuilt the whole thing, cleaning the dust as I went. Anyway, the machine started fine. Let's see if the audio difference has made any difference to the temperatures:
I class that as a complete success!
Now, while I'm here, you'll have noticed from the posts in this blog that I've been increasing the number of things I'm monitoring, which means that I'll be using more disk space in my InfluxDB server machine. Let's have a quick look at that:
Ouch! Look at that file system usage increasing over the past few months! I'm using a "forever" retention policy, so even with regular compaction built-in to InfluxDB, I'm filling my storage space. I need to change that!
Fortunately, it's one of my LXC containers, so a quick trip to Proxmox to double its size from 20GB to 40GB:
And now I'm back to having lots of disk space available:
So my InfluxData implementation is providing me lots of great insights, enabling me to see my growth over time, to understand what actions I need to perform for preventative maintenance, and how my entire infrastructure is doing.
I'm still going through pros and cons of each, so I haven't made a decision regarding the operating system to choose on the new server, but there were a couple of components I bought that aren't reliant on the new hardware, so I spend a few minutes last night performing incremental upgrades.
The new components to upgrade that server are a couple of Noctua NF-A12x15 PWM fans. I thought I'd replace the old case fan, and add a fan specifically to cool the hot-swap drive cage.
The old fan was replaced easily, with the Noctua fan using its supplied rubber mountings rather than screws. Then onto the drive cage fan - there is no mounting in my case for a fan to blow air directly across the drives. However, there are some ventilation cut-outs in the drive cage, which happen to be in roughly the right places for fan mounting, so I was able to position the other fan using the rubber mountings.
The motherboard in that NAS box is an ASRock A330ION, which is a 10-year-old design for low-power small-form-factor enthusiast server. It is equipped with 4 fan headers.The motherboard is shipped with a chipset fan, but the CPU is passively cooled.
I started the server, and a big puff of dust was expelled! Two things about this: 1) it must be more effective then before, or it wouldn't have expelled the dust. 2) Eww perhaps I should have disassembled and rebuilt the whole thing, cleaning the dust as I went. Anyway, the machine started fine. Let's see if the audio difference has made any difference to the temperatures:
Now, while I'm here, you'll have noticed from the posts in this blog that I've been increasing the number of things I'm monitoring, which means that I'll be using more disk space in my InfluxDB server machine. Let's have a quick look at that:
Fortunately, it's one of my LXC containers, so a quick trip to Proxmox to double its size from 20GB to 40GB:
So my InfluxData implementation is providing me lots of great insights, enabling me to see my growth over time, to understand what actions I need to perform for preventative maintenance, and how my entire infrastructure is doing.
Comments
Post a Comment