Starting in supernode mode

[knahrvorn]

So, if I want to run an Armory supernode, I have to run it with the --supernode flag. Do I have to do that every time I start Armory, or just the first time (generating the db)? What happens if I generate the db as a supernode and then later start Armory without the supernode flag? (or the other way around) Will the db then contain both supernode and normal entries, and would that be bad?

Do you plan to add a new start icon for supernode mode (like for offline mode)?

[goatpig]

Do I have to do that every time I start Armory

Yes

What happens if I generate the db as a supernode and then later start Armory without the supernode flag

It won't let you

[jl2012]

I'm starting in supernode mode with a virtual machine with 4GB Ram and quad-core CPU. It took about 20 hours and it's still in 51% of "building databases". It says it will take 1.5day to complete. Is this speed normal?

[goatpig]

I'm starting in supernode mode with a virtual machine with 4GB Ram and quad-core CPU. It took about 20 hours and it's still in 51% of "building databases". It says it will take 1.5day to complete. Is this speed normal?

SSD? With 4GB of RAM, in a VM, I'm not too surprised. I takes me ~4h at this blockchain size with 5820k, 32GB of RAM and SSD raid. I'm working on some changes that should speed that thing considerably (and mostly make it more stable).

[jl2012]

I'm starting in supernode mode with a virtual machine with 4GB Ram and quad-core CPU. It took about 20 hours and it's still in 51% of "building databases". It says it will take 1.5day to complete. Is this speed normal?

SSD? With 4GB of RAM, in a VM, I'm not too surprised. I takes me ~4h at this blockchain size with 5820k, 32GB of RAM and SSD raid. I'm working on some changes that should speed that thing considerably (and mostly make it more stable).

It's Raid 1 with ordinary HDD. I'm increasing it to about 6GB RAM. See if that helps.

[goatpig]

I dont expect that will be enough to be significant.

[jl2012]

I dont expect that will be enough to be significant.

Although it's still very slow, but it improves a lot (>100% improvement)

[goatpig]

Interesting, I didn't expect to react so well to so little RAM

[jl2012]

Interesting, I didn't expect to react so well to so little RAM

83% now and 14 hours for the remaining 17%

[goatpig]

Interesting, I didn't expect to react so well to so little RAM

83% now and 14 hours for the remaining 17%

The deeper you get in the more expensive each % gets, because all this data gets really fragmented within the one big DB file. Next version should help a lot with that.

If I had one tip to give you, it would be to close Armory when it's done scanning and create a copy of the DB folder. This way if it gets corrupted you can start right off of that one. Again, next version will be resilient to the kind of corruption this version suffers from.

[jl2012]

It is starting "scanning tx history" and says it will take 1.5 week to complete