dr|z3d
sure, sites that are publishing 4,0
orignal
and most i2pd publush 4 only
dr|z3d
right, which is why they won't be on identiguy.
dr|z3d
Correction, T3s|4.. when the console tab isn't visible in the browser, ie another tab is active. doesn't matter if it has focus or not, but it's worth noting that with the console tab active and the browser behind other windows, refresh will still occur.
orignal
i2pd uses reg anyway
dr|z3d
identiguy isn't authoritative, it's not a "legit" registrar, but if it wants to be taken seriously, it needs to support ECIES-only hosts.
dr|z3d
otherwise it's just another one of those abandoned sites that used to be useful.
orignal
elgamal is not but troubles
orignal
I'm going to discontinue it completely
dr|z3d
yeah, it's probably time it got deprecated.
orignal
and also risk of attacks
dr|z3d
anything new and different you've noticed re monkey attacks? LS attack still active?
orignal
I don't see any attack going on
dr|z3d
LS count normal then?
orignal
like 600-800
orignal
still lack of transit at 2RRY
dr|z3d
let me see what the status of 2RRY is on a router.
dr|z3d
interesting, just not seeing it in the netdb.
dr|z3d
not banned, either.
orignal
although it's floodfill
orignal
e.g. you sjould see it
dr|z3d
sure
dr|z3d
ok, visible on another router. not banned.
orignal
yes. but the question is why it's not visible everywhere
dr|z3d
LeaseSets: 353Routers: 19609
dr|z3d
First heard about: 70 hours agoLast heard about: 28 min agoLast heard from: 23 sec ago
orignal
Routers: 15730 Floodfills: 612 LeaseSets: 433
orignal
right now
dr|z3d
LeaseSets: 433Routers: 15611
dr|z3d
First heard about: 70 hours agoLast heard about: 32 min agoLast heard from: 11 sec ago
dr|z3d
smells about right.
orignal
yes
orignal
still no on other routers?
dr|z3d
No routers with [Hash prefix 2RRY] found in the network database.
dr|z3d
Only testing 2.
dr|z3d
Speaking of testing, has anyone tested some of these new Gen5 NVMEs?
orignal
what CPU do you need for it?
dr|z3d
for gen5 nvme?
orignal
yes
dr|z3d
something fairly recent with pcie5 nvme support.
dr|z3d
I'm looking at the speeds, ~10GB/s write, 13GB/s read.
dr|z3d
so not so much the cpu as the chipset/motherboard.
dr|z3d
I mean, 550MB/s from SATA ssds isn't exactly slow.
dr|z3d
tempted to upgrade to a mobo/cpu combo that supports gen5 nvmes, orignal? :)
orignal
yes, but you also need right CPU for it
orignal
not only mothreboard
dr|z3d
motherboard kind of dictates cpu.
dr|z3d
you'll want AM5 or the latest intel stuff.
dr|z3d
and a recent chipset on the mobo with support for native gen5 nvme.. plenty only support gen4.
dr|z3d
cpu-wise, as a developer you'll probably want 8 cores min on amd, though you could probably get away with 6.
orignal
am5 ))
orignal
most of my hardware is am4
dr|z3d
looks to be slightly more expensive, am5, but that's the price of progress (and price gouging!)
dr|z3d
apparently on AM5 48GB ram sticks are a thing.
dr|z3d
2 of those and you're good for a while.
dr|z3d
let me know if a hard refresh fixed the refresh, T3s|4__
T3s|4_
dr|z3d: as always, I run + with FF and many tabs open. After a restart, with only + in focus, Uptime remains 'frozen' at 6 seconds, then the same after hard refreshes at say: 24 s, 36 s, 58 s, and 107 s. Tunnel counts are updating despite Uptime remaining frozen
dr|z3d
if you refresh after a restart, it should rejig the refresh, T3s|4_
dr|z3d
you should only need to do that once for everything to continue refreshing. known issue, post-restart refresh needs a kick.
dr|z3d
it's on the TODO list :)
dr|z3d
let's see if we can expedite the item on the todo list :)
dr|z3d
so I've tweaked that specific element, hopefully it should now refresh after a restart. not sure why it isn't, the parent container (table) is set to refresh, but anyways, new build forthcoming.
dr|z3d
obviously it'll take a restart, hard refresh in browser and then another restart to confirm if it's fixed. let me know!
dr|z3d
more updates to snark also in the latest build.. you'll need to be running with the ubergine theme to see them all.. prolonged button depression, some visual and interaction tweaks to the snark control buttons when activated.
dr|z3d
also, we're now using variable page load buffers for the console pages, which you might notice as faster pageloads.. verdict's out, let me know...
dr|z3d
ok, new update available in 10 or so minutes, I see the issue you're reporting here. good catch!
dr|z3d
I forget that we have several options to display that info, not just one.
dr|z3d
Blinded message
dr|z3d
new update ETA 5m.
dr|z3d
sidebar refresh javascript streamlined, was 15K, now 5K, hopefully without any loss of functionality and your uptime refresh fixed, T3s|4 :)
T3s|4
Thanks dr|z3d - looks like you found it and fixed it in the first 13+ release. Just so know in the 02:51:46 build, hard refreshes had no impact getting uptime to run on its own :)
dr|z3d
thanks for testing, T3s|4. good stuff.
dr|z3d
ok, latest + dev build now comes with an inline text viewer for snark.
dr|z3d
saves the hassle of text files opening in a new tab.
dr|z3d
if you want the text in a new tab, you'll see a link icon appended to the filename.
eyedeekay
gitlab will be back in 5 minutes
zzz
dr|z3d, you're not putting new updates at skank.i2p/dev/i2pupdate.zip any more?
dr|z3d
I am, zzz, what makes you say that?
dr|z3d
Are you taking them for a spin, perchance? :)
zzz
last mod oct. 25
zzz
*26
zzz
after 301 redirect to skank.i2p/i2pupdate.zip
dr|z3d
ah. that'll be it.
dr|z3d
you need to update to the release version before you update to dev builds.
zzz
regardless, it's old
dr|z3d
saves me the hassle of having to resolve issues for those that do a cross-grade straight to dev builds and wonder why their install's hosed (no pack200 lib).
dr|z3d
do you recall when last we pushed out a release?
zzz
it's in history.txt and console news
dr|z3d
2024-10-07 2.7.0+ (API 0.9.64) released
dr|z3d
so in the scheme of things, a release build from the 25th October isn't old...
zzz
I'm on my 10th rewrite of LS expire, chasing RAP/RAR issues
StormyCloud
berry old
zzz
sure but it's not 10 seconds old given your checkin pace ))
dr|z3d
right, that'll be the /dev/ path, assuming you're running + when you update :)
dr|z3d
*** gives some time for it to sink in :) ***
dr|z3d
-rw-r--r-- 1 www-data www-data 8182275 Dec 2 10:01 i2pupdate.zip
dr|z3d
re LS expiry, are you suggesting that your previous fix isn't as robust as you thought it was?
zzz
the WIP fix wasn't tested last time we discussed it
dr|z3d
to be clear, are we talking about the same thing? the yellow status issue?
zzz
no this is ls spam mitigation
dr|z3d
ah, ok.
dr|z3d
so back to my updates, did you want a dev build url that doesn't require + to download? you never answered my question re taking it for a spin?
zzz
just had it configged in canon as unsigned update, don't remember why, not actually downloading it
zzz
still on Oct. 26 using eephead
dr|z3d
yup, it will always return the release version unless you're running +
zzz
you're changing the HEAD results based on user-agent?
dr|z3d
no
zzz
or, you're redirecting or not based on user-agent I guess
dr|z3d
UA-based redirect.
dr|z3d
<dr|z3d> saves me the hassle of having to resolve issues for those that do a cross-grade straight to dev builds and wonder why their install's hosed (no pack200 lib).
zzz
got it
zzz
back to my RAP/RAR rabbit hole
dr|z3d
what's your current WIP strategy?
zzz
as discussed a few days ago, aggressive expire if over a limt
zzz
limit
dr|z3d
ok
dr|z3d
something akin to the scaling expiry for RIs?
zzz
yerp
dr|z3d
sounds good
zzz
but I have to unhork RAP/RAR first
dr|z3d
sounds like a party I'm happy I'm not attending :)
Quaddle
Hmm.
Quaddle
How does it work. Taking the pgp route on vpn ?
dr|z3d
?
Quaddle
Tor and the art of anonymity says : If you're going to go the VPN route, then use PGP: Pretty Good Privacy.
Quaddle
Never, ever transmit plain data over a VPN, not even one that offers SSL.
dr|z3d
Same with Tor.
dr|z3d
only more so.
Quaddle
But how do i fit a pgp key into vpn ?
dr|z3d
they're probably referring to mail.
Quaddle
Ah, yes. That could be.
zzz
success, crawling out of the RAP/RAR hole
zzz
mostly unfinished business from subdbs that needed to be cleaned up
onon_
I may not be able to participate in the discussion tomorrow. So here is my (very important) opinion:
onon_
You will not succeed.
onon_
Because if the intermediate node is not required to reserve bandwidth, then it is useless
onon_
And if it is required, then it creates new attack vectors.
zzz
thanks for your comments
zzz
it's required to reserve bandwidth if it says it did, although it can and probably should overcommit
zzz
what attack vectors are you thinking?
onon_
As an attacker, I can request an infinite number of tunnels and not use them.
onon_
And regular legitimate users will not get their bandwidth
zzz
well, we don't accept an infinite number of requests, we will start rejecting
zzz
wouldn't a resonable overcommit strategy be sufficient?
zzz
'legit' requests would either get accepted or rejected, if accepted they would get bw, if not they would build a different tunnel
onon_
I will send requests constantly in large quantities, and legitimate users - few.
onon_
And my requests will be accepted and legitimate ones - rejected
zzz
agreed, it's a real concern
zzz
but couldn't we track total requested vs. actual bw, and overcommit on that basis?
onon_
In that case, it would mean that we are not reserving anything, if I understood your idea correctly.
onon_
And this, as I already said, is no different from the current situation.
zzz
well, it is 'reserve', but somewhere less than a guarantee, because we can't predict 10 minute into the future. Our local bw could increase, leaving less for part. traffic
zzz
and, of course, the requestor's indication of minimum bw req'd is just an estimate too
onon_
As I already said, the positive effect is highly questionable, but the vector for attack is quite real.
zzz
yes, I appreciate the feedback. The proposal lays out the protocol enhancements necessary to implement it, but the implementation itself and nuances like overcommitment are not specified
zzz
it seems like you're saying there's no foreseeable implementation that would not be attackable or at least undesirable?
onon_
I couldn't come up with one like that.
zzz
an 'attack' where the only consequence is that tunnels get built through a different router isn't much of an attack, but network-wide it could be an issue
zzz
my intent was to specify a protocol so that we could start implementation, and explore these issues, perhaps there's no valid solution as you say
zzz
if you're not there tomorrow I'll do my best to represent your concerns
zzz
thanks again
onon_
I'm glad if I could be of some help
zzz
always happy to have eyeballs on things, doesn't happen too often ))