This is good idea. Thank you!
That’s why I pay for an extra internet connection, just for kids… and a node 
.
With chinese IoT is verry probable to get a cryptominer or become a bot for hackers. I received as bonus a cryptominer in an official firmware update for Popcorn Hour 4k media player 
. It had a verry powerful cpu at that time. It was activated only when I put the player in stanby mode. I realised something is not right because it was getting verry hot in standby, and some parts of the GUI didn’t work after the update… and my ISP blocked me one day for suspect traffic. I didn’t had the skills to analise traffic, and did’t bothered with it. Just switched it off and sold it on sh market.
We also have two internet connections here. But for work reasons rather than for the kids. Still I was extremely happy to get the VLAN separation in place.
I’ve heard the horror stories. When the wife wanted to get TV streaming - told her that’s fine but don’t trust the crappy hardware. Heck, I don’t even trust the tp-link AP’s so using them for an untrusted VLAN makes perfect sense,
Update: it’s gone. Traffic returns to normal now.
As expected. These customers seems stopped/finished upload.
Yes, you may configure pfSense to be your router or run just a pure Linux VM and configure it to be your router.
Even Windows is some kind of capable (but you must not disable a firewall).
In my travels, I sometimes used my Windows laptop as a wireless access point for my other devices if Internet access was only available through an Ethernet cable (yes, I carried that cable with me too). Even in this configuration it was better than most of consumer routers.
Any particular model of this Enterprise SSD ? Can’t find anything for such a low price.
samsung 840 pro? maybe?
not a enterprise ssd.
2 Likes
I just ordered a pair of these INTEL SSD DC S3500 2.5" 120GB Laptop SATA Solid State Drive SSDSC2BB120G4 | eBay for a new setup I’m putting together. 120GB Intel DC S3500, for $10. The dude even gave me another 10% off.
And here is a datasheet: https://www.intel.com/content/dam/www/public/us/en/documents/product-specifications/ssd-dc-s3500-spec.pdf#page14
Generally, you would want the seller to post SMART attributes to judge the remaining endurance — but at this price and for this purpose it’s absolutely irrelevant. Even if there is just 10% life remaining — it shall be fine.
At 10% life remaining this doesn’t look like a good deal. You can now get a new 120GB drive with a similar TBW rating for around 25 USD.
2 Likes
2 SSD for l2arch in raid0?
Yes, the point was, for this usecase it’s fine. No reason to spend $25 when $10 suffices; and I do agree with you that this is poor value.
Plus, likely they are not actually 90% used up. I’ve asked seller, he said “he does not know”. The worst I saw from sellers with such plausible deniability was 25% endurance left. Do I expect 10%, because magic does not exist; I’ll let you guys know once I receive them :).
At this price it does not matter, nor you need it for intended use. The write speed to metadata device is quite low to begin with; then exhausting the endurance rating does not turn it into pumpkin; you just have reduced power-off retention (order of months); but this device is going to be powered 24/7, so it does not matter. Once the endurance is 10x exceeded, they may start failing as readonly, and ZFS will start sending metadata back to the array.
Basically, I would use the cheapest SSD there (but not consumer one: you want to have PLP and firmware to fail gracefully, into readonly mode, as opposed to dropping off the bus entirely),
No. Mirror them and add as a special device, to hold metadata and small files.
L2ARC is not very effective with storagenode. It will accelerate second and subsequent accesses, special device accelerate including the first. Node seems to read a small subset of data repeatedly and the much larger chunk randomly or never.
4 Likes
I can tell from my setup, 55%-60% of ~25GB cached per TB node data are second and subsequent accessed. databases are already excluded.
as this is my constant cache hit ratio. (after an analyse with ultradefrag to beginn with)
so i conclude 500 gb readcache are enough for an 20TB node full of data.
I really don’t know how many nodes some people are running behind one router. But I’m running at least 7 of them, not having any router-related troubles. Ingress sometimes over 300GB/day, without any issues. And yeah, it’s an ASUS. So what? Before it was a ZTE (ISP-choosen) router, also up to the task. So, people plagued with choking and smoking routers, I’m really wondering whether they are running their node over WiFi or something.
But I’m living in The Netherlands, with almost everyone having the option of 1Gbps internet access which is becoming the default here. Although, some people are still bound to a meager 100Mbps. Might make a difference in standard quality of default routers too…
You don’t want WiFi in the first place, but if you’re bound to… Such like with your mobile phone and so on 
Well, I don’t like the attitude. I mean, the idea of spreading it the most, could be very well served by a lot of home-hosting SNO’s. Especially, since there was a period that even a Raspberry Pi 3B (1GB RAM) was sufficient according to the handbook. Probably it should be taken into account, that there are different kind of SNO’s.
So the software should be serving them all, not exactly the same way. For example rate limiting download speeds or checking RAM usage / caching in order to prevent choking SMR-drives. Till then, limiting the amount of upload orders is a good proxy.
Exactly what I mean…
No, it was a Raspberry 3B. See for example Install storagenode on Raspberry Pi3 or higher version – Storj.
And actually, my nodes (even the SMR ones) all do well with 1.5GB of RAM. Most of them using only several 100’s of MB’s.
3-7W on average, still worth it in almost every country I hope?
Mind the fact, that the node is in another pool for the first 100 audits (I believe) for every satellite, since it needs to be vetted first. So, it’s an option that takes some time to help you out.
Well, that was a jackpot: 97% and 96% life remaining on the two drives I ordered. Probably these were from corporate laptops.
1 Like
it doesn’t exclude running operation from home,
just approaching it more seriously, and get more care from STORJ inc.
Yea “diffrent types of SNOs”, we shall not bother, coz STORJ inc. is turning passion to commercial SNOs i think. as its less demanding, and more profitable.
Public network of SNOs like us, may not exist in 2 years if they score great success with ComSNOs with Certs, so why bother with classic SNOs improovments?..
You are wrong. These operators for SOC2 compliance, not to compete with the public network.
We need all kind of operators, preferably - in a different geographical locations, like
1 Like
Oh im very wrong. You don’t get what i am saying.
Not now, but in general STORJ business model.
COO already was in UK’s parliament and memo describes STORj only, as using existed space in data centers, according to X twitt.
" John Gleeson, COO, Chief Operating Officer for Storj said: “The greenest data centre is the one that’s never built or expanded. Our technology leverages existing, unused storage capacity in tens of thousands of data centres all over the world. We reduce the carbon footprint of cloud storage by 60-80% primarily by avoiding the need to purchase new hard drives and extending the service life of the drives already in use."
It’s a shift of focus, so yea, tell me again how wrong i am. Thx.
1 Like