Bitcoin full node and query server

November 2017

Archive of the libbitcoin Slack #general channel.

Wednesday, November 1, 2017

neillm00:36
i'll start with mainnet2, it's building now and going down shortly
evoskuil00:41
Hold on, I have another thing I want to patch which will requires resync. We are failing to properly categorize P2SH spends because the pattern matching conflicts with other spend types.
neillm00:41
ah, ok
evoskuil01:13
GitHub
I have tried to get the balance of a P2SH address by using blockchain_fetch_history3(), but the returned value is not what I expected. It will return the value of all of the transactions related to...
neillm01:32
nice, good finds

Thursday, November 2, 2017

evoskuil07:46
This is now complete in v3 and master. I still have to port the former issue to master, but v3 (head) is ready for deployment testing.
I'll be in Palo Alto for Scaling Bitcoin from Friday through Monday morning.
evoskuil11:13
Master is now updated as well. However note that v3 is the release quality branch - master is staging for v4.
neillm22:57
mainnet2 testnet going down shortly for an upgade and re-sync
i mean testnet2 of course

Friday, November 3, 2017

evoskuil08:53
I'll be in Riga, Latvia for Baltic Honeybadger 2017, Nov 25/26
evoskuil09:59
master and version3 branches are now both updated. There was an additional change in libbitcoin, so if you had already updated server, sorry - need to do it again as of just a minute ago. This is basically two patches that both affect server indexing of address history. P2SH indexing was largely missed, and there was a race that caused infrequent dropping of entries for an addresses with multiple entries in multiple txs of one block. Once the community servers are updated and we have good feedback I'll tag a v3.4. Thanks everyone for the reporting and help!

Sunday, November 5, 2017

neillm19:03
mainnet2 and testnet2 going down shortly for the recent v3 update and re-sync

Saturday, November 11, 2017

neillm04:02
mainnet2 is finally back(!). testnet2 has been back for some time
mainnet1/testnet1 are next up. testnet1 first though
evoskuil10:55
:)
neillm15:27
hrm, testnet2 w/o checkpoints seems to be getting stuck after block 11100 still
the mainnet2 syncs were with checkpoints, and those completed
mainnet2/testnet2, that is
Rejected block [0000000028502288da2fa584d24403536a3755351fd53a39f8e9d48d170
9fc0e] from [107.182.230.232:18333] block hash rejected by checkpoint
and sorry, i mean this is testnet1 (i can't type this am, must be confusing ;-)
in short, mainnet2/testnet2 completed with checkpoints. built fresh v3 for testnet1, and no checkpoints has an error after block 11100
neillm19:34
for now i'm going to enable checkpoints for testnet1
evoskuil19:43
What is the error in the log when it gets stuck?
neillm19:46
the one pasted above, rejected block
evoskuil19:47
Did you verify the checkpoint?
Or are you saying it is rejecting for checkpoints despite having none?
If so you probably have default checkpoints and that is a mainnet point.
You need to keep the genesis block checkpoint in order to prevent default behavior.
neillm19:50
ah, let me try that
actually, that seems to have fixed it
i didn't realize that a single genesis checkpoint was needed. I mostly only sync with all checkpoints, to the latest at the time of config
trying to sync with none, seemed like i could just comment them all out. incorrect :-)
evoskuil19:55
Well default behavior always takes hold in that case.
neillm19:56
and default is mainnet genesis?
evoskuil19:56
Yes. I may eventually eliminate default checkpoints in v4.
neillm19:57
i think it's fair, i just didn't realize. maybe worth warning that the node is configured for testnet but is using a mainnet checkpoint? not sure
glad that it's user error in any case. instead mainnet1 sync, i am planning to scp the store from mainnet2. if we need to test a full mainnet sync w/o checkpoints, i will keep mainnet1 running and re-sync mainnet2. otherwise, after the scp, both will remain up and i'll just update when the tags come in
evoskuil20:02
Mainnet should be verified without checkpoints as well. The 3.4 changes affect address indexing, but I may have a consensus bug to patch when I get home in a few days. First one in a couple years, but not positive yet.
If so I will patch it in 3.4 as well.
neillm20:04
so the current status is this: mainnet2/testnet2 and now testnet1 are the very latest heads of version3. mainnet1 is the original v3.3. the plan was to use the mainnet2 store (i believe it's 3.4 compatible) on mainnet1
thus, the scp. after mainnet1 is up with the 3.4 compatible store, i can erase mainnet2 and re-sync with full validation
unless the new bug invalidates the store and everything needs to be re-sync'd :-)
neillm20:39
so that's the plan anyway at the moment. keep me posted and i'll rebuild when tagged

Sunday, November 12, 2017

evoskuil01:36
It won't invalidate the store, but it would be nice to get a full sync. It may be sufficient to checkpoint the full sync though, since the check is block static and prob wouldn't be skipped by checkpoints.
neillm04:28
testnet1 back online and sync'd
@evoskuil either way, just let me know. the main gripe is i want maximum uptime for mainnet1/testnet1 since tests during builds attempt to hit them and i want to minimize errors due to downtime
neillm19:07
as a point of reference, the store on mainnet1 (vanilla v3.3) is currently 312GB. The store on mainnet2 (pre-tagged v3.4) is currently 276GB. a solid ~11.5% reduction.
evoskuil19:19
Yeah, it was just wasted space.
(edited)

Monday, November 13, 2017

neillm03:08
both mainnet1 and mainnet2 will be going down shortly and will be down until the updated store is copied over (estimating an hour or so).
Updated estimate: 2 hours
neillm04:51
mainnet1/mainnet2 back online!
evoskuil04:52
Thanks!
neillm04:53
np. still need to resync mainnet2 w/o checkpoints at some point, but crashing now ;-)

Wednesday, November 15, 2017

evoskuil21:27
I've patched this issue in master and version3: https://github.com/libbitcoin/libbitcoin/issues/834
GitHub
During code demo with me @jameshilliard noticed that we may be missing a context-free (check) block consensus rule. Need to confirm the rule existence before implementing. Rule may be the consequen...
This is the addition of a missing context-free block consensus check.
This does not affect the store (unless of course you have synced an invalid block, though its unlikely many miners would build on it).
evoskuil21:35
There is no need to resync community servers as I am resyncing locally and will verify the chain against the servers.
evoskuil23:20
This patch is now merged.

Thursday, November 16, 2017

neillm01:58
community servers updated. was the version tagged?
evoskuil02:17
Not yet.
neillm02:23
ok good, wanted to make sure i didn't mess up :-)

Friday, November 17, 2017

neillm14:36
Mainnet1/mainnet2 may be down. Power outage. Can be resolved later, but if a resync is needed, it may be a week or so before they're back in working order
evoskuil21:02
Yes, they are both down.
I'll reconfigure DNS for the default names to point to mainnet3/testnet3. However the testnet3 public query interface is not responding (secure interface is, so it's probably misconfigured).

Saturday, November 18, 2017

neillm02:14
ok, thanks, but ugh
they are both definitely down and currently re-syncing from scratch
testnets should be available later tonight
neillm05:25
testnet1/testnet2 back online!
evoskuil06:47
👍

Sunday, November 19, 2017

evoskuil01:30
I've redirected testnet.libbitcoin.net to testnet1 due to an apparent configuration problem with testnet3, which is currently responding on secure queries only.
Due to mainnet1/2 being behind from the power outage/resync, mainnet.libbitcoin.net is currently directed at mainnet3.
So both default addresses are fully operational. Thanks maintainers for the support!

Monday, November 27, 2017

skaht19:24
What time slots that caught your fancy?
evoskuil21:14
Nothing in particular. There are 4 panels and I speak in 1/2/4. Libbitcoin is mentioned several times.

Tuesday, November 28, 2017

neillm04:06
thanks for the youtube links!
evoskuil04:53
👍

Thursday, November 30, 2017

evoskuil19:26
Regtest support is now fully implemented in the version3 branch.