-
mawk
I installed libreadline-dev but the cmake says it doesn't find it
-
mawk
suspicious
-
mawk
ah I dropped the cmake cache and now it found it
-
mawk
bad bad cmake
-
ndorf
yeah :\
-
mawk
I suppose BUILD_GUI_DEPS is what I need to build the GUI?
-
ndorf
sounds plausible. i've never built the GUI though, sorry
-
mawk
I'll give the cli wallet a try
-
mawk
the GUI makes my fan spin a lot for some reason, probably related to the deluge of weird errors in the log
-
ndorf
that would be expected while syncing, but does it happen after it's fully up to date, too?
-
mawk
even from a remote node it would be syncing with lots of CPU usage?
-
ndorf
yes,
-
mawk
well it's fully up to date and every now and then it spins up
-
mawk
for the fan to spin up the cpu usage of a few cores must be 100% for some time
-
ndorf
a local node uses CPU while syncing because it needs to verify the incoming blocks,
-
ndorf
a wallet (regardless of local or remote node) also uses CPU while syncing, because it needs to *scan* all the outputs in the new blocks
-
mawk
right
-
ndorf
so if you use a remote node, you don't need the first half, but you always need the second half no matter what
-
mawk
yeah
-
mawk
unless it's the so called lightweight wallet I suppose
-
ndorf
"every now and then" spinning up might just be when a new block first arrives?
-
mawk
which doesn't provide much security
-
ndorf
right, with a lightweight wallet the server does the scanning
-
mawk
yes I suppose that's the case
-
mawk
my raspberry pi has a ssd, maybe I could try putting the node there if its cpu would be good enough
-
ndorf
it's just as secure, the server never has access to your spend key. what you give up is privacy, since the server can see all your incoming transactions.
-
mawk
ah right yeah
-
ndorf
which model pi? i use a rockpro64 for a monero node and it works great. i'd guess rpi4 would work just as well
-
mawk
it's a rpi4 yeah
-
mawk
already serving as a router, wifi AP and NAS
-
mawk
but none of these are CPU intensive
-
mawk
and it has a fan but I unplugged it because the noie was obnoxious, so there will be definitely no fan noise to annoy me there
-
ndorf
i don't have a fan on my rockpro, just the passive sink. doesn't seem to have any issues.
-
ndorf
it didn't seem to have issues when syncing from scratch either, more pertinently
-
ndorf
i think running it 24/7 on a cheap ARM board like this is a pretty good choice. personally i'd be a little wary of running anything open to the internet on the same machine as my private NAS, however
-
mawk
yeah indeed
-
mawk
it's the router so it has a public IP
-
ndorf
if you are confident you can isolate it properly (e.g. use firejail), it's probably fine, but i prefer to just spend the $50 and get a second one
-
ndorf
public IP is one thing, but something actually accepting connections is another
-
mawk
but there's some firewall on it which should be good enough, unless monerod gets pwned then it's not good enough
-
mawk
for now it doesn't accept ingoing connections, and I'd just add the P2P ports of monerod and that should be good
-
mawk
the IP isn't static though, I don't know how much a problem it can be
-
ndorf
when you add the p2p port, monerod will accept connections on that port
-
mawk
yeah
-
mawk
other nodes will know my ip, and just mark it as stale when they can't connect anymore?
-
ndorf
yeah, dynamic IP is not a problem, i would bet the majority of nodes are on one
-
mawk
I see
-
mawk
if you believe in my abilities to do a proper firewall, you think it's fine to run monerod on the pi like this? or you would still be wary of a vulnerability in it and put it in a secure container thing
-
mawk
putting it in an unprivileged container would be an easy way to isolate it
-
mawk
and with systemd it's easy to do in a few lines of config
-
ndorf
i don't think the firewall makes any difference since the relevant port would be open anyway
-
ndorf
depends how sensitive the stuff on your NAS is? if it's just movies you downloaded then whatever :)
-
mawk
yeah just movies
-
mawk
but the pi is also the router
-
mawk
if it's a skilled attacker they could try to access my other computers with more sensitive stuff on it
-
ndorf
otherwise, at a minimum, i'd make sure monerod has its own unprivileged user that doesn't have access to anything (e.g. your NAS files)
-
mawk
as computers tend to trust their LAN and especially their router
-
mawk
yeah
-
ndorf
you can also add a firewall rule preventing that user from sending anything to your internal network
-
mawk
ah yes good point
-
mawk
unless it's RELATED,ESTABLISHED traffic
-
ndorf
beyond that, i'd personally set it up under firejail, it's pretty easy to do
-
ndorf
exactly
-
mawk
I've looked at firejail before it looks powerful indeed
-
mawk
but apart from the more exotic seccomp capabilities of firejail a container should be more or less the same result
-
mawk
firejail might be easier to use tho
-
mawk
thanks for the advice
-
ndorf
yeah, you also don't need a whole separate chroot system with firejail. instead it provides a limited view of the host filesystem
-
ndorf
so in this case you'd just give it read-only access to e.g. /usr/lib/whatever and /usr/local/bin, and write access to wherever your blockchain and log file are
-
mawk
yeah
-
mawk
maybe a stupid question but the p2p protocol is udp right?
-
mawk
I suppose
-
mawk
to get through NAT
-
mawk
is there a way to type a command into a remote node ndorf ? if it's unrestricted RPC for instance I should be allowed right?
-
mawk
otherwise I would have to like login on the pi and restart monerod in interactive mode to type the command
-
mawk
estimated time for syncing 2 days, with a ssd :(
-
mawk
I'll just let it run and see what happens
-
mawk
but it told me I'm on a rotating drive which is false
-
mawk
ah now it's 22 hours estimated
-
mawk
my own estimate would be 3-6h given the current speed
-
nioc
It slows down later in the chain due to larger blocks/more txs later
-
nioc
Also the first 40% of the chain was a different and faster format
-
nioc
Not sure but a sata ssd might take 22hrs
-
nioc
An nvme will be ~12hrs
-
ndorf
mawk: you can send commands to a remote node like this: monerod --rpc-bind-ip node.whatever.com flush_txpool
-
ndorf
--rpc-bind-port X if it's not 18081
-
saunders
are the large exchanges any worse, or better, than the smaller monero specific exchanges when trading fiat for monero?
-
moneromooo
TCP
-
noex
so i'm trying to create a node, and it's been syncing for a few days...but it oscillates between syncing normally, and throwing stack traces. is that...normal behavior?
-
noex
after Exception: it prints one of those diamond question mark characters which i think is a non-printable character
-
noex
and then ultimately: src/common/stack_trace.cpp:149 Failed to find the next frame
-
noex
but it seems to recover after a while
-
noex
whatever is happening appears to be a recoverable error
-
noex
but it will print pages and pages of stack traces before resuming normal sycning
-
noex
Synced 2309456/2526760 (91%, 217304 left, 2% of total synced <--- this has also been confusing me
-
noex
it appears to be 91% synced, and the other "total synced" number almost seems randomly generated. it jumps all over the place.
-
moneromooo
It's meant to be "total synced since you last started syncing again", while the 91% is the total from chain start.
-
moneromooo
If it jumps, possibly you're disconnecting all the time and it's restarting sync.
-
moneromooo
For the stack traces, they should appear in the log file only. Are you seeing them on the console ?
-
noex
yeah
-
moneromooo
Very odd. Did you compile monerod yourself ?
-
noex
but i'm doing tail -f logfile :P
-
moneromooo
...
-
moneromooo
OK, I *was* imprecise, technically :)
-
moneromooo
Can you paste a sample stack on paste.debian.net ? It's most likely something uninteresting.
-
moneromooo
You can get rid of them with --log-level 0,*stack*:FATAL
-
noex
-
moneromooo
Whee. Well, not helpful. I guess you did not biuld with libunwind.
-
moneromooo
But most likely innocuous anyway :)
-
moneromooo
The two things it's likely to be are:
-
moneromooo
- a bad weak_ptr in the net code: I'd fixed that (mostly) but it came back, but it's really innocuous, there's a check for it, it's just a noisy way to check whether an object has died
-
moneromooo
- failure from librandomx to use huge pages or the like: it'll fall back on normal allocs, it'll just be a bit slower
-
moneromooo
If it's another than these, it'd be interesting to know, but little chance of that.
-
noex
yeah it seems innocuous because it looks bad, but I will check later and it's syncing totally normal. output is fine.
-
noex
just keeps chugging along
-
noex
if it's only supposed to appear in the logfile and not regular output, that makes me feel better. maybe next time I won't detach and run it in tmux or something
-
noex
i guess i thought it was just standard output
-
ocb
h
-
gingeropolous
any other small box recommendations? ima get rpi4, i have a pine64 from years ago...
-
ocb
beaglebone
-
ocb
wanted to get beagleboard-x15
-
hyc
rockpro64
-
hyc
avoid all rpi models
-
ocb
hyc: mind elaborating?
-
hyc
they all lack hardware AES
-
hyc
they all have shitty slow I/O subsystems
-
ocb
i will look into it, as im planning to get one more SBC soon and was rethinking what.
-
ocb
thanks
-
hyc
20MB/s max throughput, aggregate of all interfaces
-
hyc
waste of perfectly good silicon
-
ndorf
even the pi 4 still lacks AES? damn
-
fluffypony
yeah
-
fluffypony
no clue why they don't use processors with crypto extensions
-
fluffypony
so stupid
-
ndorf
almost but not quite unbelievable
-
hyc
cheapskates didn't want to pay for the extension license
-
rbrunner
That would be a license paid to ARM? AES itself does not require anything, right?
-
hyc
from broadcom to arm, yes
-
ndorf
wtf. i have a different ARM SBC (NanoPi A64) that was $15 new and it has AES
-
hyc
broadcom makes all the pi CPU chips. either raspberry specifically asked for this, or this is what broadcom offered them and they took it
-
hyc
afaik broadcom makes the chips to raspberry's specification