00:14 |
asciilifeform |
!w poll |
00:14 |
watchglass |
Polling 6 nodes... |
00:14 |
watchglass |
205.134.172.27:8333 : Alive: (0.078s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618884 (Operator: asciilifeform) |
00:14 |
watchglass |
205.134.172.4:8333 : Alive: (0.132s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=618884 |
00:14 |
watchglass |
205.134.172.6:8333 : Alive: (0.142s) V=50400 (/"therealbitcoin":0.5.4/) Jumpers=0x1 (TRB-Compat.) Blocks=618884 |
00:15 |
watchglass |
108.31.170.3:8333 : Alive: (0.357s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618884 (Operator: asciilifeform) |
00:15 |
watchglass |
188.121.168.69:8333 : Alive: (0.261s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618884 |
00:15 |
watchglass |
103.36.92.112:8333 : Busy? (No answer in 15 sec.) |
00:15 |
asciilifeform |
!w probe 103.36.92.112 |
00:16 |
watchglass |
103.36.92.112:8333 : Alive: (0.971s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618884 |
| |
~ 1 hours 27 minutes ~ |
01:43 |
feedbot |
http://bvt-trace.net/2020/02/a-tiny-and-incomplete-trb-wedgetrace/ << bvt's backtrace -- A tiny and incomplete TRB wedgetrace |
| |
~ 7 hours 11 minutes ~ |
08:54 |
asciilifeform |
!w poll |
08:54 |
watchglass |
Polling 6 nodes... |
08:54 |
watchglass |
205.134.172.6:8333 : Alive: (0.082s) V=50400 (/"therealbitcoin":0.5.4/) Jumpers=0x1 (TRB-Compat.) Blocks=618944 |
08:54 |
watchglass |
205.134.172.27:8333 : Alive: (0.084s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618944 (Operator: asciilifeform) |
08:54 |
watchglass |
108.31.170.3:8333 : Alive: (0.099s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618944 (Operator: asciilifeform) |
08:54 |
watchglass |
103.36.92.112:8333 : Alive: (0.527s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618944 |
08:54 |
watchglass |
188.121.168.69:8333 : Alive: (0.180s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618944 |
08:54 |
watchglass |
205.134.172.4:8333 : Busy? (No answer in 15 sec.) |
| |
~ 1 hours 11 minutes ~ |
10:06 |
asciilifeform |
!w poll |
10:06 |
watchglass |
Polling 6 nodes... |
10:06 |
watchglass |
205.134.172.27:8333 : Alive: (0.025s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618947 (Operator: asciilifeform) |
10:06 |
watchglass |
205.134.172.6:8333 : Alive: (0.082s) V=50400 (/"therealbitcoin":0.5.4/) Jumpers=0x1 (TRB-Compat.) Blocks=618947 |
10:06 |
watchglass |
108.31.170.3:8333 : Alive: (0.188s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618947 (Operator: asciilifeform) |
10:06 |
watchglass |
103.36.92.112:8333 : Alive: (0.527s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618947 |
10:06 |
watchglass |
188.121.168.69:8333 : Alive: (0.334s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618947 |
10:06 |
watchglass |
205.134.172.4:8333 : Busy? (No answer in 15 sec.) |
| |
~ 23 minutes ~ |
10:29 |
asciilifeform |
!w probe 205.134.172.4 |
10:29 |
watchglass |
205.134.172.4:8333 : Busy? (No answer in 15 sec.) |
10:29 |
asciilifeform |
sumbody's wedged... |
| |
~ 1 hours 16 minutes ~ |
11:46 |
BingoBoingo |
asciilifeform: ty, looking at top before restarting the explosion in RAM usage suggests it may be the phenomenon bvt documented |
11:52 |
asciilifeform |
BingoBoingo: i'm going through his piece atm actually. dollars to doughnuts, it's underflow in unsigned int nSize = vSend.size() - nMessageStart; but atm can't prove it |
11:54 |
asciilifeform |
the proggy i'm aboutta genesis might prove useful in replicating this effect (it so happens to contain working routines for injecting arbitrary liquishit into a noad) |
11:55 |
asciilifeform |
BingoBoingo: i take it .4 is yours btw ? wouldja like it labeled ? ( so can see yer name highlighted when probed ) |
11:56 |
* |
asciilifeform has of course list of who in cage is who, but did promise not to advertise folx w/out their req |
11:56 |
BingoBoingo |
Maybe later |
11:56 |
asciilifeform |
aite |
11:57 |
shinohai |
mmmm doughnuts. |
11:57 |
asciilifeform |
shinohai: 4GB shitdoughnut, lol |
11:58 |
asciilifeform |
will be interesting if it turns out that that 4GB is actually entire memory space of process and gets sent out to issuer of 'magic cmd' |
| |
~ 1 hours 11 minutes ~ |
13:10 |
feedbot |
http://www.loper-os.org/?p=3665 << Loper OS -- "Watchglass" Genesis. |
13:20 |
asciilifeform |
!w poll |
13:20 |
watchglass |
Polling 6 nodes... |
13:20 |
watchglass |
205.134.172.6:8333 : Alive: (0.022s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618965 |
13:20 |
watchglass |
205.134.172.27:8333 : Alive: (0.144s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618965 (Operator: asciilifeform) |
13:20 |
watchglass |
108.31.170.3:8333 : Alive: (0.097s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618965 (Operator: asciilifeform) |
13:20 |
watchglass |
103.36.92.112:8333 : Alive: (0.527s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618965 |
13:20 |
watchglass |
188.121.168.69:8333 : Alive: (0.458s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=618965 |
13:20 |
watchglass |
205.134.172.4:8333 : Busy? (No answer in 15 sec.) |
13:32 |
* |
asciilifeform reground the vpatch to put proggy in named dir, as is trad. |
| |
~ 32 minutes ~ |
14:05 |
bvt |
asciilifeform: nah, nSize looks correct, and pend in |
14:06 |
bvt |
sorry, fatfingered. actually, nSize would look ok after such overflow, so it may just be masking the problem |
14:09 |
asciilifeform |
bvt: as i understand, for this effect, either a) vSend.size() b) nMessageStart c) vSend.begin() d) vSend.end() wildly outta bounds, or d) memory corruption boojum somewhere entirely else |
14:09 |
asciilifeform |
*e) |
14:13 |
bvt |
these 4Gb of memory look like a realloc result to me, caused by an underflow somewhere in serialization code |
14:14 |
asciilifeform |
bvt: if can find just where, i can write a liquishit sender so to reproduce. |
14:20 |
asciilifeform |
bvt: for completeness -- could also be locking bugola: there's a number of places where e.g. nMessageStart = -1 |
14:22 |
bvt |
another datapoint is that it happens when a node requests a ton of blocks, apparently could be an under/overflow if vSend.size() becomes bigger than fits in uint32? |
14:24 |
asciilifeform |
that's what suggested 'race cond.' to me -- that it happens , apparently , strictly under max cpu load |
14:27 |
asciilifeform |
astonishingly gnarly shitcoad, incidentally. and nobody's touched the 'message' crapola since genesis . |
14:27 |
bvt |
hm; i could wrap operator<< of vSend to check how the size() changed - if by more than 4Gb (3Gb for some slack?), we have a culprit? |
14:28 |
asciilifeform |
bvt: of course can 'kludge' , but won't know whether nailed it until effect can be reproduced |
14:29 |
asciilifeform |
bvt: i dun suppose you have a core dump of wedged state ? |
14:30 |
bvt |
i have it in debugger in wedged state right now, so i'd like to get a maximum value out of it before killing. |
14:31 |
asciilifeform |
at the very least, make dump if you can |
14:32 |
asciilifeform |
bvt: can get values of the stack vars of EndMessage at the time Hash invoked ? |
14:32 |
bvt |
ok, have 5.4Gb core.pid |
14:33 |
bvt |
info locals? |
14:33 |
asciilifeform |
plz |
14:35 |
asciilifeform |
bvt: i'ma set up a place you can put this turd, gimme coupla min |
14:36 |
bvt |
nothing interesting http://bvt-trace.net/2020/02/a-tiny-and-incomplete-trb-wedgetrace/#comment-135 (I'll use comments because pastes tend to disappear) |
| |
~ 19 minutes ~ |
14:55 |
asciilifeform |
bvt: http://paste.deedbot.org/?id=Me42 |
14:58 |
asciilifeform |
bvt: lemme know plox if this worx. |
15:11 |
bvt |
should finish uploading the dump in ~15min |
15:11 |
asciilifeform |
ty bvt ! |
15:11 |
asciilifeform |
dun fughet the orig bin too plz |
15:12 |
asciilifeform |
bvt: i'ma gzip & put it in pub in case others want to look |
15:26 |
bvt |
i'm not sure this is a good idea, given that i'm not sure what amount of information about the host may end up in the dump. |
15:26 |
asciilifeform |
bvt: your call |
15:27 |
asciilifeform |
bvt: whatever's in there, fwiw, already came outta yer box , i strongly suspect, when it went an' tried sending out that 4gb |
15:28 |
asciilifeform |
at the very least, oughta give mod6 a copy (not sure whether he knows how to gdb tho?) |
15:28 |
bvt |
should be it |
15:29 |
asciilifeform |
ty bvt |
15:29 |
asciilifeform |
imho would be valuable to put it up publicly. if you have nuffin seekrit on that box aside from pw's, i rec to change'em, then can put up ? |
| |
~ 26 minutes ~ |
15:56 |
asciilifeform |
bvt: ? |
15:57 |
bvt |
if BingoBoingo acks, i'll make it available via my blog, you can share the data as well |
15:57 |
asciilifeform |
ok i'ma mirror on my box then |
15:57 |
asciilifeform |
(plenty of bw to spare) |
15:57 |
asciilifeform |
will be up in half hr or so |
15:58 |
asciilifeform |
just nao finished gzippin' |
16:00 |
* |
asciilifeform forgot that bvt were living in BingoBoingo's box |
16:03 |
asciilifeform |
bvt: moving it b/w machines atm, <15m left |
| |
~ 20 minutes ~ |
16:23 |
asciilifeform |
bvt: http://nosuchlabs.com/pub/bvt-trb-wedge.tar.gz |
16:23 |
asciilifeform |
sha512==bebfd877d7bc1e8790d6dba1a0d41b0fa591884bd07c3b08ed55b1ae87f470a707c168bafe7c070e38a9a85c762a02b3361c07eb56bc981ecffa86940ef9aa16 |
16:28 |
asciilifeform |
( of 'core.2266' : ae3eeac1e13f01a5c4f9979498ff367e4571f89e32ec1ac4d78d59428bf48fc13d750e93c83ddb92796178759b54e919c4e7f6a336364fe8fe3213b11f1c2b07 ; |
16:29 |
asciilifeform |
of 'bitcoind-wedge2' : 793e8fe8d37620b6f9b6cb14e9e40e04cb45f1e9bce3ff3785a3ed427c455f1442bd35b17e1fd6746a6e321154ea5c81ab439e545dcd32a91cfcad52bb38bb02 . ) |
16:33 |
feedbot |
http://blog.mod6.net/2020/02/multiple-vpatches-tested-and-signed/ << mod6's Blog -- Multiple Vpatches Tested And Signed |
16:43 |
asciilifeform |
bvt: i put the mirror link in your comments (atm in spam queue) . |
| |
~ 51 minutes ~ |
17:35 |
shinohai |
lol was trying mod6's new patchset for trb in above blogpost, phexdigit fix causes it to fail on esthlos-v: http://btc.info.gf/paste/9a33aa@raw |
| |
~ 25 minutes ~ |
18:00 |
feedbot |
http://qntra.net/2020/02/man-arrested-after-failing-to-destroy-motor-vehicle-in-pentagon-parking-lot/ << Qntra -- Man Arrested After Failing To Destroy Motor Vehicle In Pentagon Parking Lot |
| |
~ 24 minutes ~ |
18:25 |
shinohai |
^ above mystery solved, mod6 wrote in just as I was tracing problem and evidently I was missing some previous patches. |
18:25 |
shinohai |
I only have up to 'aggressive_pushgetblocks' in current patchset, will sign a mirror to www as soon as all tests pass. |
| |
~ 23 minutes ~ |
18:49 |
feedbot |
http://fixpoint.welshcomputing.com/2020/bitcoin-transactions-and-their-signing-1/ << Fixpoint -- Bitcoin transactions and their signing, 1 |
18:54 |
shinohai |
BingoBoingo: possible lulzy Qntra story, I simply don't have time to write up this evening: https://archive.is/aLNq0 |
18:54 |
shinohai |
Run moar SehShit |
18:54 |
shinohai |
*SegShit |
18:55 |
BingoBoingo |
tyvm |
18:56 |
asciilifeform |
bvt et al : http://bvt-trace.net/2020/02/a-tiny-and-incomplete-trb-wedgetrace/#comment-137 |
19:09 |
asciilifeform |
... continued . |
| |
~ 43 minutes ~ |
19:53 |
Vexual |
interesting new tool release asciilifeform |
19:54 |
asciilifeform |
Vexual: by all rights farthest thing from 'interesting'. but potentially useful imho. |
20:02 |
Vexual |
potentially useful is inherently interesting |
20:04 |
asciilifeform |
Vexual: what i meant was, there's nuffin nontrivial in there. ( ate 1 afternoon. ) ircism lifted from ye olde logotron bot, plus bit of logic to ask 'version' packet from noades & report answr. |
20:06 |
Vexual |
I assumed you already had it; and irc bit was new |
20:06 |
asciilifeform |
Vexual: had irc bit (since august), the rest new |
20:07 |
Vexual |
Well then I was almost precisely wrong |
20:08 |
asciilifeform |
imho ideally it'd probe in background (say, erry 20min) and report strictly wedged and 'behind' nodez |
20:08 |
asciilifeform |
but would rather atm have the manually-cranked tester than nuffin |
20:15 |
asciilifeform |
orig. started w/ aim to make injector for replaying mod6's packet capture into a node. there, ended up stuck, but realized can quickly make general-purpose node poll proggy from the pieces, for the time being. |
20:18 |
Vexual |
does a testchain fit anywhere in hour head? |
20:18 |
Vexual |
it's possibly illegal by now |
20:19 |
asciilifeform |
Vexual: other than for playing w/ mining logic, wat good does 'test chain' do ? |
20:19 |
Vexual |
mining is how I got there; I've been pondering your 0 fee double spend question for days |
20:20 |
asciilifeform |
Vexual: you can make 'test chain' any time you like, just by setting up coupla nodes on a closed lan & plugging'em into ea. other |
20:21 |
Vexual |
yeah, I aslo kinda assumed it might be useful for repeating wedge conditions |
20:21 |
asciilifeform |
you'll have a hell of a time making the sort of block fillage & mempool volume seen in 'nature' tho |
20:21 |
asciilifeform |
so doubtful re wedge conditions |
20:22 |
Vexual |
copy nature |
20:22 |
asciilifeform |
( this is even supposing 'wedge packet' is 'natural', i.e. accidental, rather than crafted, object ) |
20:23 |
Vexual |
good luck testing on mainchain |
20:23 |
Vexual |
test yes, repeat no |
20:23 |
asciilifeform |
Vexual: re 'doublespends' -- pocket test network of trb won't tell you much re the current miners, who quite evidently use very diff. logic than seen in trb's (which has ~0 optimization for tx selection) |
20:27 |
Vexual |
I agree |
20:28 |
asciilifeform |
re wedge -- given that apparently 0 prb nodes suffered the wedge, i'm inclined to think that this is the answer. even tho atm cannot readily say how to replicate condition. |
20:30 |
asciilifeform |
tomorrow will take deeper look at bvt's dump, possibly derive the necessary input to replicate effect. (supposing no one does 1st while i sleep) |
20:33 |
asciilifeform |
!w poll |
20:33 |
watchglass |
Polling 0 nodes... |
20:33 |
asciilifeform |
hrm 1s i think i moved config, lol |
20:37 |
asciilifeform |
!w poll |
20:37 |
watchglass |
Polling 6 nodes... |
20:37 |
watchglass |
205.134.172.27:8333 : Alive: (0.084s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619004 (Operator: asciilifeform) |
20:37 |
watchglass |
205.134.172.4:8333 : Alive: (0.083s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=619004 |
20:37 |
watchglass |
205.134.172.6:8333 : Alive: (0.082s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619004 |
20:37 |
watchglass |
108.31.170.3:8333 : Alive: (0.104s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619004 (Operator: asciilifeform) |
20:37 |
watchglass |
103.36.92.112:8333 : Alive: (0.589s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619004 |
20:38 |
Vexual |
ssalghctaw pals\ |
20:38 |
watchglass |
188.121.168.69:8333 : Busy? (No answer in 20 sec.) |
20:39 |
asciilifeform |
!w src |
20:39 |
watchglass |
asciilifeform: my source code can be seen at: http://www.loper-os.org/?p=3665 |
20:39 |
asciilifeform |
there. |
20:39 |
asciilifeform |
oughta be ok to leave alone nao. |
20:39 |
Vexual |
:) |
20:46 |
asciilifeform |
Vexual: have you a node ? it is a little odd that yer interested in subj, but no noad ? |
20:47 |
Vexual |
nope, my strange node sits on a platter collecting dust |
20:48 |
Vexual |
prolly get hot'n'loud if I dusted it off |
20:48 |
asciilifeform |
my e.g. 'apu1', w/ 0 moving parts , makes an entirely workable noad. |
20:49 |
asciilifeform |
is approx the size of 2 vhs cassettes laid long sides together |
20:50 |
asciilifeform |
draws <10w |
20:51 |
asciilifeform |
that's what lives at .27 |
20:52 |
Vexual |
with halvening and covid19, I might just need to spin something upi |
20:53 |
asciilifeform |
to paraphrase the old song, 'i roll 2 nodes in 2 times of peace, and 2 in times of war, i roll 2 nodes before i roll 2 nodes, and then i roll 2 moar' |
20:53 |
asciilifeform |
or how it went. |
20:53 |
Vexual |
not sure if soviet or jamaican |
20:54 |
asciilifeform |
was engl. so prolly the latter |
20:57 |
asciilifeform |
!w poll |
20:57 |
watchglass |
Polling 6 nodes... |
20:57 |
watchglass |
205.134.172.6:8333 : Alive: (0.082s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619005 |
20:57 |
watchglass |
205.134.172.4:8333 : Alive: (0.083s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=619005 |
20:57 |
watchglass |
205.134.172.27:8333 : Alive: (0.084s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619005 (Operator: asciilifeform) |
20:57 |
watchglass |
108.31.170.3:8333 : Alive: (0.160s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619005 (Operator: asciilifeform) |
20:57 |
watchglass |
103.36.92.112:8333 : Alive: (0.467s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619005 |
20:57 |
watchglass |
188.121.168.69:8333 : Alive: (0.343s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=619005 |
20:58 |
asciilifeform |
ha, all 6. |
20:58 |
* |
asciilifeform bbl. |