-
Steven_M
Hi all. Is the fork late?
-
selsta
the fork happened already
-
selsta
there will be a second fork tomorrow
-
DeanGuss
in monerod.conf I have: p2p-bind-port=18080 and p2p-bind-ip=127.0.0.1 and my onion points at blah.onion:18080 -> 127.0.0.1:18080
-
DeanGuss
I also have anonymous-inbound=blah.onion:18083,127.0.0.1:18083,64
-
DeanGuss
Are they both necessary?
-
DeanGuss
I also have the same blah.onion pushing 18083 and 18081 (rpc) connections to 127.0.0.1
-
DeanGuss
I gather that remote rpc is required for me (or others) to use my node as a remote node - is that true?
-
hyc
yes
-
hyc
I wouldn't publish port 18081 though
-
hyc
I'd only publish the restricted-rpc port
-
DeanGuss
That is my restricted rpc port
-
DeanGuss
but ok so ignoring 18081 for now, why are there 2 different things to do the same thing? Are they just 2 different ways of saying the same thing (considering how I have my p2p-bind-port and p2p-bind-ip set?
-
hyc
what two different things?
-
DeanGuss
p2p-bind-port and p2p-bind-ip, along with anonymous-inbound. Last I checked I couldn't put them on the same port
-
hyc
sure you can
-
hyc
aren't you already putting them on 18080?
-
DeanGuss
I'm putting p2p-bind-ip on 18080 and I am putting anonymous-inbound on 18083 as suggested when that config option came out
-
DeanGuss
iirc I wasn't able to put them on the same port. Glad to hear I can, I'll try it
-
hyc
ahh
-
hyc
yeah I have mine on 18083 as well
-
DeanGuss
FATAL net contrib/epee/include/net/abstract_tcp_server2.inl:1277 Error starting server: Failed to bind IPv4 (set to required)
-
DeanGuss
ERROR daemon src/daemon/main.cpp:364 Exception in main! Failed to initialize p2p server.
-
hyc
yeah they have to be separate
-
hyc
p2p-bind-ip / p2p-bind-port is only for clearnet
-
hyc
any other network (tor, i2p) requires its own port
-
selsta
you don't anything onion on 18080
-
DeanGuss
selsta: why not?
-
DeanGuss
anonymous-inbound=blah.onion:18080,127.0.0.1:18083,64
-
DeanGuss
^ I just did that and it seems to have worked... at least externally. I had to go with 127.0.0.1:18083 internally there
-
hyc
yeah that can work, because it's really using 18083 on the local machine
-
hyc
the outside world talks to onion:18080 but tor forwards it to local:18083
-
hyc
on one of my nodes I'm using onion:18083 -> local:18084 because zmq-pub is already on local:18083
-
DeanGuss
So if I totally leave out anonymous-inbound=... if I were to connect to say another .onion node, would it see my IP as a tor exit node's public ip? And anonymous-inbound makes it so that they see my .onion? Or how does it go?
-
hyc
it means what it says
-
hyc
anonymous-inbound only sets how other peers can connect to you
-
hyc
has no effect on how you connect to them
-
DeanGuss
How do they learn that .onion name?
-
hyc
it's published in peerlists like everything else
-
DeanGuss
K. And if I had no anonymous-inbound option, and just used a hidden service and p2p-bind-ip=127.0.0.1 and p2p-bind-port=18080, what would they see then?
-
hyc
so it only can get published because you first contacted a clearnet peer. like one of the seed nodes
-
hyc
monerod doesn't know anything about your hidden service unless you tell it via anonymous-inbound
-
hyc
if you don't use that flag, nobody tries to contact you there.
-
DeanGuss
OK. Do they try to contact me at all or can I only do outbound with that config?
-
hyc
clearnet ...
-
DeanGuss
woah... so you're saying I'm connecting to outbound peers via clearnet?
-
hyc
yes
-
DeanGuss
even with p2p-bind-ip=127.0.0.1 ?
-
hyc
bind-ip only affects incoming connections
-
hyc
has no impact on outbound connections
-
DeanGuss
wow so... I *still* have to use torify if I want monero to use tor?
-
DeanGuss
after all this time
-
hyc
you use --tx-proxy to tell it to use tor outbound
-
DeanGuss
that's only for when I submit transactions to peers I thought
-
hyc
but monerod always uses clearnet
-
hyc
if you want everything outbound to go thru tor then yeah, use torsocks or whatever
-
DeanGuss
my jaw is on the floor right now
-
DeanGuss
how can that still be true
-
hyc
?
-
DeanGuss
that we still can't proxy outbound connections via tor somehow within monero
-
hyc
in general, tor is too slow
-
DeanGuss
I don't buy that. It's fine. All my bitcoin traffic goes through tor just fine. It's not like monero has more transactions than bitcoin
-
hyc
there's also the problem of fending off sybil attacks if everything is going thru tor
-
DeanGuss
Not if you use tor exit -> clearnet
-
DeanGuss
I'm not trying to only use .onion
-
DeanGuss
just I need everythign going through tor and actually
-
DeanGuss
I think it already does that
-
selsta
DeanGuss: --proxy flag
-
DeanGuss
I just did a netstat and grepped for monero and I see there is not a single clearnet connection. Everything is to 127.0.0.1:9050
-
DeanGuss
selsta: hrmm... I seem to recall that, but thought it was fazed out when anonymous-inbound and tx-proxy came around
-
selsta
no
-
hyc
so you're already using tx-proxy
-
selsta
but --proxy will use exit nodes
-
DeanGuss
hyc: yes of course
-
selsta
since we don't support p2p via onion, only transactions
-
DeanGuss
does tx-proxy apply to every tx that I relay as well as submit?
-
hyc
yes
-
selsta
relay from other nodes?
-
DeanGuss
selsta: yes
-
hyc
there'd be no point if the txs were treated differently, that'd make them traceable
-
DeanGuss
hyc well if that's true why would I need torsocks?
-
selsta
quite sure they got propagated to clearnet
-
hyc
sounds like you don't need torsocks, I prob misspoke earlier
-
hyc
didn't understand your setup
-
DeanGuss
and what is the difference between --proxy and -tx-proxy ? --proxy is for outbound fetching of txs/blocks?
-
selsta
proxy is a normal proxy, it does not understand tor
-
selsta
tx-proxy has specific tor / i2p support and is specifically for tx
-
DeanGuss
What do you mean understand tor? You mean get .onion transport/name resolution?
-
hyc
help for tx-proxy says it only affects local txs. didn't notice that before
-
selsta
it will use the tor exit nodes, it does not connect to onion p2p nodes
-
DeanGuss
selsta: Why would it care? Does it not implement socks4a/socks5h (where the proxy does the name resolution)?
-
hyc
but my daemons routinely log "unable to send tx to tor" and I haven't sent a txn t that time. I suppose it's relaying via dandelion+
-
selsta
hyc: i think there might be 1 hop before they get to clearnet
-
selsta
DeanGuss: what do you want to do exactly?
-
DeanGuss
selsta: I want to advertise to the network that they can connect p2p for blocks and transactions at my onion, I want to connect to .onion and clearnet ips (but only via tor), and I want to submit transactions via tor. I want everything to go out and come in over tor and I want clearnet in the mix to protect against sybils.
-
DeanGuss
and I want restricted rpc available over my .onion as well
-
hyc
you can't say "everything go out and come in over tor" *and* "clearnet in the mix"
-
DeanGuss
yes I can, why not?
-
hyc
to keep clearnet in the mix, something has to go out and come in over it
-
DeanGuss
I want to reach clearnet via tor
-
DeanGuss
not just .onion
-
hyc
ah
-
hyc
then we're back to using --proxy as selsta mentioned
-
hyc
no idea how that looks in the peerlist
-
selsta
like I said before, you can't sync blocks over onion, it will use exit nodes to connect to other clearnet nodes
-
selsta
.onion nodes are only supported with --tx-proxy
-
Wallet
selsta: Your default coin is now set to XMR. Change with coins command.
-
Wallet
selsta: ≈$0.0588 • ≈ value of: 1 ONION • Source: cmc/ccc/altm
-
selsta
you can use both --proxy and --tx-proxy
-
hyc
what inbound addr:port is advertised to the peerlist when you use --proxy?
-
selsta
the one from the exit node I would guess
-
selsta
and inbound port is always random from what I remember
-
hyc
monerod creates the peerlist entry, not the proxy. and the proxy isn't bidirectional
-
hyc
monerod doesn't know its exit node address
-
DeanGuss
selsta: so --proxy is for proxying to clearnet and --tx-proxy is for proxying to .onion, but I can't fetch whole blocks from .onion nodes? Or do you just mean I can't fetch from .onion nodes with --proxy?
-
selsta
yes
-
selsta
I meant yes to the first
-
selsta
there are no .onion pees with --proxy
-
selsta
peers
-
DeanGuss
hyc: I dunno a log about monero (so thank you and selsta for explaining at length), but if you have `bitcoin-cli getpeerinfo` handy it might be worth looking into how they do it: For each peer they list eg. "network": "ipv4", "addr": "ip_of_peer:8333", "addrbind": "127.0.0.1:port", "addrlocal": "tor_exit_ip:port" where I have bitcoin setup to always use tor and connect to .onion and any ipv4 ip via tor.
-
DeanGuss
s/log/lot/
-
DeanGuss
selsta: ok cool. So it sounds like I can achieve what I want if I use --proxy and --tx-proxy (everything goes through tor and I can talk to .onion and clearnet via tor)
-
selsta
yes
-
DeanGuss
Thanks.
-
DeanGuss
So is there a concept of a reindex in monero? What if something goes wrong with the database index?
-
DeanGuss
Maybe the need for such rpc calls is obviated by lmdb somehow
-
Steven_M
selsta: oh, sorry. I didn't know about the 2nd fork.
-
Steven_M
selsta: Is 0.18.0.0-release sufficient for the next fork?
-
DeanGuss
Steven_M: Looks like yes unless you're using trezor or ledger with monerod:
getmonero.org/2022/08/11/monero-0.18.1.0-released.html
-
Steven_M
DeanGuss: that's what I thought, but figured I'd double check. Thanks. :-)
-
Mochi101
Almost up to 9 hours here selsta and am at 81%
-
Mochi101
about 1/2 a million blocks to go
-
Steven_M
So why are there two hard forks so close together instead of a combined one?
-
Mochi101
To make sure the network clears of any old style txs
-
Mochi101
befire totally blocking them
-
Mochi101
It's way overkill... but this is Monero. ;)
-
Steven_M
Mochi101: Monero, finishing the job that Bitcoin left half-done. ;-)
-
Mochi101
yes
-
solinux_
Hello everyone
-
rbrunner
Hello. "Everyone" is still a bit thin on the ground on this quiet Sunday :)
-
solinux
rbrunner: he he :)
-
hyc
DeanGuss: nothing can go wring with an LMDB index
-
hyc
wrong*
-
DeanGuss
cool, figured
-
hyc
\bitcoin needs that because the DB it uses doesn't do atomic transactions
-
hyc
so it can get interrupted in the middle of an update and leave stale index info
-
moneromooo
< hyc> DeanGuss: nothing can go wring with an LMDB index <-- windows can
-
moneromooo
IIRC it was found to reorder some writes ?
-
hyc
reordering doesn't matter as long as fsync happens when intended
-
hyc
in safe:sync mode
-
moneromooo
Except if it crashes in the middle. Which it does.
-
moneromooo
And I suspect there's still a bug somewhere in db_lmdb.cpp, there's just too much report of corruption.
-
hyc
hmm. I have an upstream patch for Windows to do unbuffered writes, that may solve any reordering issue
-
hyc
corruption only on windows?
-
hyc
I've never had any corruption on my nodes but they're all linux
-
hyc
might try this commit da0527ac75b811419b7007202799f96b2edb5aef
-
moneromooo
Mostly windows at least. Never had any either.
-
hyc
and the one after it 147582b5dcd86d1d92bb9a3fda9eb640300ccae3
-
moneromooo
Which repo is that exactly ?
-
hyc
openldap and lmdb repos
-
hyc
-
hyc
might need the off_t patches right after taht too, don't remember
-
hyc
-
hyc
doing write-thru writes means we don't even need to do an explicit flush/fsync
-
moneromooo
ty
-
moneromooo
Most windows users probably can't apply but hey, I'll mention them anyway.
-
hyc
might be worth doing an experimental build
-
hyc
but I don't have any machines that would run windows longterm
-
selsta
-
selsta
hyc: can you look up if the off_t patches are required and link to them?
-
selsta
then i can make windows binaries and i'm sure there will be a windows user who can test them
-
Mochi101
Still syncing.. 92%
-
nioc
Go mochi go
-
Torr
Mochi101: Should change the nick to "Mochi92".
-
Mochi101
A little over 20 hours
-
Mochi101
selsta, you synced in 8 hours?
-
selsta
yep
-
Mochi101
Which ssd?
-
selsta
Windows probably slows you down
-
Mochi101
I have an 860 EVO in this.
-
Mochi101
So I should upgrade to Gentoo?
-
selsta
Mochi101: so it probably is a combination of your SSD not being the fastest and Windows
-
Mochi101
Old hardware...
-
selsta
the Mac Mini I used has a fast CPU and NVMe SSD, so basically ideal conditions
-
Mochi101
I have an NVMe in my other laptop.
-
hyc
selsta: yeah seems like it. basically use all of the commits from April 24-25 2020
-
hyc
-
selsta
hyc: will try that. what would happen if we update our lmdb to master?
-
selsta
do we have any monero specific patches?
-
hyc
no, there are no monero-specific patches, everything in our tree originally came from mdb.master
-
hyc
we could prob just merge current mdb.master, I did a few full resyncs of that tree in the past
-
selsta
hyc: opened this now but will need testing first
monero-project/monero #8501
-
hyc
hmmmm. maybe. looks like b6420e12a967b8e96fd87a86dd8d105962bfdf8b
-
hyc
is missing from upstream
-
selsta
what patch is this?
-
Mochi101
If you can get me a Win binary with that patch I can sync and see if there are any performance gains.
-
Mochi101
Once this current sync is done I'll have something to compare it to.
-
hyc
that was one mooo made to tweak errcodes
-
hyc
and looking over the history, there are a few commits in our tree that came from mdb.master3 - the rawpart stuff, specifically
-
hyc
that are deleted in your PR
-
hyc
I've pushed mooo's patch upstream just now
-
selsta
ok, will readd both
-
Mochi101
selsta, didn't bother you the other day some housekeeping and all the requests to close on Github did it?
-
selsta
Mochi101: github housekeeping is always good
-
selsta
hyc: should the patches apply cleanly from master3 to master?
-
selsta
Mochi101: if you see something to close you can comment or PM me the issue
-
hyc
selsta: I don't recall now, they might have had some conflicts to resolve. best just to re-add the patches from our tree
-
hyc
or I suppose I can push those into mdb.master. I really was only intending for bugfixes and not new features to go into mdb.master now tho
-
selsta
hyc: maybe easier if I close my PR and you do it locally? then you don't have to push to master
-
hyc
ah, just create a new local branch to PR?
-
hyc
sure
-
hyc
so yeah the question is whether this gives a noticeable perf boost on windows.
-
hyc
it should...
-
hyc
and should also fix write ordering
-
selsta
Mochi101: did sync finish for you?
-
selsta
then you can try the new one
-
Mochi101
123910 blocks to go
-
selsta
will also try sync from scratch with it on Mac
-
Mochi101
Will be a nice test.
-
selsta
starting sync now
-
hyc
Mac behavior should be unchanged
-
Mochi101
Daemon is telling me that I have just over 3 hours left to sync over here.
-
Mochi101
Going on 25 hours.
-
selsta
Mochi101: mine is 49% already
-
Mochi101
The first half goes super quick. Lots of empty blocks.
-
Mochi101
;)
-
garth
-
garth
1in2 out tx
-
selsta
-
selsta
yea the blockchain would be over 1TB without bulletproofs
-
hyc
they were a miracle, badly needed