May 26, 2019

I usually start a full testnet sync in the evening and it's done by the next day. Is the limiting factor at the moment the bandwidth available or disk? A month seems rather excessive.

This does assume using an ssd -- I haven't tested on spinning disk in some time, but used to do that regularly in the v2 days.

Weird, I am on an SSD with a laptop

It seems limiting factor in this case is the disk

I've had laptops with rather slow SSDs, likely slower bus speeds. But a month still sounds quite unreasonable for testnet. Mainnet in a bad case shouldn't take longer than that 🙂

v3 or v4? Shouldn't matter at that kind of timeframe, but just curious

v3

Ok, and still having the shutdown issue between attempts?

v4 has a problem for me where it doesn't stop. I leave it for several hours and it's just hanging

yep

That's unfortunate. The only difference on my current setup is that I'm running a modified protocol and server, based on these PRs: https://github.com/libbitcoin/libbitcoin-protocol/pull/202

It's possible there is web code causing a hang that was resolved in there a while ago, but I don't recall anything like that.

That's a good PR

Thanks! It's been useful for testing here. Still some unknowns on how to handle the default content, but it's a start

It's good websockets have been added, maybe even more useful than ZMQ. Still it's a shame ZMQ itself didn't add websockets as a protocol option. It's been a lingering request for years.

Interesting, I didn't know that was on the table.

Agree though, that ZMQ is the better way to go generally, but these days, it seems almost everyone knows websockets via browser js or node.js

In any case, you're welcome to give those branches a test build and see if it resolves the start/stop issue. I'm fairly certain when I re-build vanilla v4 master, it doesn't hang, but it's been a while. I have a pile of outstanding PRs I've been testing with lately

my plan is to get a server and set it up for testing

then others can login too

The matching server PR is here: https://github.com/libbitcoin/libbitcoin-server/pull/508

Ah, something like that could be good, or chaotic 😉

Oh, about the hang -- static or shared library build?

shared, I built it locally

manually myself

I've seen a hang using a shared library build because of something related to boost on ubuntu that I was unable to resolve. Statically linking to boost worked on that machine, but I couldn't determine why it was hanging inside of boost otherwise

Gdb would consistently trap it at the same spot in some innocuous boost code on shutdown.

when I get time I will try to look where it's hanging

Thanks, am curious if that has come back

i'm on arch

I don't know that it was specific to ubuntu, but a user reported it on ubuntu and I setup a VM for testing, and was able to replicate it

there should be a stop() call in the blockchain anyway

(if that's what's causing a slow shutdown)

Is it hanging around there? https://github.com/libbitcoin/libbitcoin-blockchain/blob/master/src/interface/block_chain.cpp#L856

no idea I didn't check

ah ok there is a stop() method, right

If the shutdown hang isn't like that odd boost one I was seeing, I am curious if there's a locking issue on shutdown that I'm just getting lucky on and not seeing

I almost always build statically these days too, so am curious if that will help for now. If you get a chance, give that a shot