Rebuild data after disk fail

Hi,
My disk is died. Does someone know how to rebuild and get the data that I used to have? Just attach the new disk and start the the docker service with the same script and identity?

There is no way getting that data back. You have to get new identity and start new node. Same script should work if only disk is changed.

If you are lucky you can try a SATA to USB adapter to boot it. If that doesn’t work you’d need to send it in for recovery, but likely not worth saving the node at this point.

And this is why you should use RAID :).

Depends on the RAID level. RAID1 is not really worth it for Storj; you’ll get more revenue running two nodes, one using each disk.

Using RAID5 with many disks might make sense financially.

RAID1 may not be worth it, I agree. However, no RAID has more problems. In case of a drive failure, not only you lose escrow (which may be quite a lot, depending on when the drive failed), the new node:

  1. Gets very little traffic for a month (vetting).
  2. Has no data initially
  3. Because of no data, it gets less traffic until it accumulates enough data
  4. Escrow is back to high percentage
  5. Reputation is low, so the node may get less new data.

It all depends on how long each drive lasts and their capacities.

If we assume 1TB drives and each drive lasts at least two years, then you’re probably going to make more money running a node per disk, since the drive is probably going to fill up within the first few months. Then you are making twice the income (without RAID) from that point forward as compared to running RAID1.

RAID1 might save you some hassle (not having to create a new identity and wait for an auth token) but IMO it will generate less revenue than using both disks to run separate nodes.

If you’re using 12TB disks that last four months, the calculation will look very different.

1 Like

According to the Storj documentation:
https://documentation.storj.io/dependencies/identity

“Make a backup of your identity folder, including all 6 files contained in it, and preferrably on an external storage medium. This will allow you to restore your node to working order in case of an unfortunate incident such as a hard drive crash.”

So, this is why I asked the question! Somehow it is possible, but I did not find anything…

It is possible if you get ALL the data off your crashed hard disk. But if you’re missing data you might get disqualified soon.

:frowning: Sorry to hear that. Then I should get a new Node Identity, am I right? I just reused the same Identity and I am getting data so far…

This identity will be disqualified sooner or later on all satellites. Please, create a new identity and sign it with a new authorization token.

TBH, I never understood this advice. If the hard drive storing the node’s data dies, a backup of the identity will not do you any good. This advice only makes sense if the identity and node data are stored on different disks.

… which is quite common, at least on a linux system

2 Likes

This is a typical setup. The default path for identity is a system disk. Usually the disk for data is a separate disk.
Even for Windows it is true too

2 Likes

I guess it depends on where you’re storing your data. In my case storagenode is running in a VM and the data is on the same volume as the identity.

It might be good to clarify in the documentation under what circumstances it’s useful to make a backup. Many SNOs may think that they can recover from a failed data disk using this identity when that’s not the case.