IPFire + Storj as "hospice" for retired drives?

Node Operators getting started
1

I am currently running IPFire as a firewall/router/NAS/QoS gizmo at home. Very modest setup on very old HW, Atom330 board with 4 SATA channels and 4GB RAM (3.5GB usable).

I migrated a 1.4TB node (on a 1.5TB HDD) to it, and my barebones Storj setup seems to be working OK on bare metal. I am running the binaries directly from CLI, since there isn’t even a docker package available for IPFire.

Repurposing elderly but perfectly functional HDDs for Storj on said setup makes some sense to me, since it is running 24/7 anyway (minus the occasional IPFire upgrade), and Storj payouts are (barely) enough to cover the additional electricity cost of spinning extra drives.

All “green” and fine AFAICT, but does such recycling make much sense from Storj’s perspective? The drive(s) would be relatively slow, and my upload speed isn’t great either. Only upside is, what with QoS, I should be able to guarantee the minimum 5Mbps upload requirement at all times.

I do have a bit more powerful bits of retired HW handy, but if this “hospice” concept of mine is outright undesirable for Storj even as a redundancy, I might just as well do a graceful exit.

1 Like
2

If it makes sense for you, financially, then run it. Recycling old useless hardware is a good idea, in fact none of my PCs are younger than 10 years old and the disks are used but only 3-5 years, except I bought a new 20TB disk just for storj a while back

Edit : huh, that new 20TB disk was in Nov 2022, so not so new anymore lol

1 Like
4

For a large number of smaller drives like that the only idea that resonates in me is to buy an open docking station and use them as if they were floppies or tapes: removable storage, normally not even powered. There’s plenty of software that can help organizing content. Keeping them on 24/7 is a waste of electricity.

For a single drive or so, there’s usually that old desktop for retrogaming you want resurrected. It will not complain that the disk is small or slow. I was giving away small drives to help friends do so.

1 Like
5

I had a box of old laptop drives and a guy who rebuilt old school xboxes bought them for his work. Life… finds a way…

My electricity is expensive, so I won’t even spend the 5W or 10W for storj hosting unless the drive is big enough to make the payoff worth it.

tangentially, how do you like ipfire? Do you have previous experience with something like pfsense to compare it against?

6

Fortunately, my electricity is dirt cheap for now, cheap enough to run old HW like I mentioned before. Speaking of old stuff, my (also Linux-based) VDSL2 modem/firewall/router/wifi box is no longer eligible for upgrades, so I beefed up my security by dumbing the box down to what amounts to a modem/switch/wifi AP, and set up IPFire for security.

Unsurprisingly, I narrowed my choices down to pfsense and ipfire, then chose ipfire because I have a little bit of Linux experience but big fat zero FreeBSD experience. I am no security expert, nor do I feel qualified to judge between those two, all I can say is that so far (~1 year) I haven’t had any particular urge to try pfsense just for the sake of trying.

The obvious aside, dynamic DNS works, Samba works, QoS works, I even managed to use it as a VPN server… barely. The lowly AT3IONT-I Deluxe just doesn’t seem to pack enough oomph to yield satisfactory VPN bandwidth. Anyway, all that from browser GUI, SSH CLI is for funky stuff like Storj. Barebones, haven’t figured out how to set up Storj as a service, but I don’t mind starting Storj manually, once in every blue moon or so.

Apparently there is one funny side to this particular setup tho:

IMG_0421
IMG_04211928×1408 290 KB

No (half-baked) performance enhancements, no fancy-pancy attack vectors… :woozy_face:

1 Like
7

And the saga continues… the initial setup is back to doing just the firewall thing and samba shares for my home network; that’s what it’s quite good for.

I migrated my 1400GB node to somewhat mightier piece of junk, a 4C/4T i5 with 8GB RAM Ubuntu server. Put in 2x 1.82TiB HDDs and a 56GiB SATA SSD to experiment with ZFS.

Still bare bones, / and /home are on SSD, and I still have identity, databases and orders under /home/storj user. Still starting node manually. The newness is ZFS metadata on SSD as well - 42GiB partition - and a small log partition just for the heck of it. Not so sure if Storj benefits from the log, though.

jml@storjnode:~$ sudo zpool list -v ; echo "" ; sudo zpool iostat -vl
NAME        SIZE  ALLOC   FREE  CKPOINT  EXPANDSZ   FRAG    CAP  DEDUP    HEALTH  ALTROOT
storj      3.67T  1.41T  2.26T        -         -     0%    38%  1.00x    ONLINE  -
  sda      1.82T   716G  1.11T        -         -     0%  38.6%      -    ONLINE
  sdb      1.82T   718G  1.11T        -         -     0%  38.7%      -    ONLINE
special        -      -      -        -         -      -      -      -         -
  sdc3       42G  9.10G  32.4G        -         -    39%  21.9%      -    ONLINE
logs           -      -      -        -         -      -      -      -         -
  sdc4     4.84G      0  4.50G        -         -     0%  0.00%      -    ONLINE

              capacity     operations     bandwidth    total_wait     disk_wait    syncq_wait    asyncq_wait  scrub   trim  rebuild
pool        alloc   free   read  write   read  write   read  write   read  write   read  write   read  write   wait   wait   wait
----------  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----
storj       1.41T  2.26T     78     28   904K  1.04M    2ms    2ms  863us  789us  200us    1us    6ms    2ms    5ms  798us      -
  sda        716G  1.11T      1      2   109K   390K    9ms    2ms    9ms    1ms    2us    1us    1ms    2ms      -      -      -
  sdb        718G  1.11T      1      2   129K   391K    8ms    3ms    8ms    1ms    2us    1us   46us    2ms      -      -      -
special         -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -
  sdc3      9.10G  32.4G     76     22   666K   285K    1ms    2ms  598us  634us  208us    1us    6ms    2ms    5ms  798us      -
logs            -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -      -
  sdc4          0  4.50G      0      0    104    211    2ms  455us    2ms  455us  493ns    1us      -      -      -    1ms      -
----------  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----  -----
jml@storjnode:~$

Obviously this looks pretty overkill and silly, zero metadata redundancy and a striped VDEV does not exactly spell reliable long-term operation. Then again, it still is an experiment/marginally useful hospice for old hardware. Bright side is, previously vigorous HDD thrashing seems to have quieted down very nicely. The new setup has remained practically silent even with just a modest CPU cooler and one case fan, so I presume it isn’t an outrageous electricity hog either.

Starting to dream bigger… IPFire does have a qemu addon, so with better HW (and more memory) I might try to run, say, TrueNAS in a VM under IPFire host. Just one consolidated chunk of hardware doing firewall, home NAS, storj node, plus cloud backups to Storj…

Sounds fascinating that TrueNAS apparently has both Storj client and node add-ons baked in. Use some, rent more, and have the Storj node cover the cloud backup cost. Woooo…

Obviously much more “pro” setup, I am starting to entertain thoughts of 2xSSD for OS and mirrored metadata, plus 3x modern (big) HDDs in ZFS RAIDz for home NAS and Storj node datasets. Not going to happen in the near future though, need more means and skills to try and pull that one off. However, it does sound like something that could work, doesn’t it?

Core 180 update virsh qemu passtrough harddisk and truenas - Add-Ons - IPFire Community