11:12 |
awt |
$ticker btc usd |
11:12 |
busybot |
Current BTC price in USD: $22285.68 |
| |
~ 2 hours 10 minutes ~ |
13:22 |
asciilifeform |
awt: i'm on the new blatta, and seeing megatonnes of warning barf |
13:23 |
asciilifeform |
awt: "Peer with duplicate address or handle id detected: shinohai" |
13:23 |
asciilifeform |
kilometres of this |
13:31 |
asciilifeform |
!q seen-anywhere shinohai |
13:31 |
dulapbot |
shinohai last seen in #asciilifeform on 2022-08-29 09:21:10: lol juche no worries I was intentionally being facetious since had never seen you here before and only saw above prose. ^_^ |
13:31 |
asciilifeform |
!q seen shinohai |
13:32 |
dulapbot |
shinohai last seen here on 2021-11-18 09:48:13: that *is* cool lel |
13:32 |
asciilifeform |
hm bot bug? |
13:33 |
asciilifeform |
( or not? ) |
13:33 |
asciilifeform |
also ftr seeing the multi-minute message embargo thing again |
13:37 |
asciilifeform |
btw awt -- imho the 'send addrcast with 1/60 odds erry min.' oughta be per cold peer, rather than all in 1 burst |
13:38 |
asciilifeform |
( the way it is now, is too easy for snoop to calculate how many cold peers a station has from traffic burst mass ) |
13:40 |
shinohai |
asciilifeform: when I updated to 9973 I mysteriously had peers not in user list, though their keys are present when I run %wot |
13:41 |
shinohai |
I will try to upgrade to latest blatta and figure this out when I get settled here - Castle Titsbare was revealed to have problems with chinesium mold so am temporarily mobile. |
13:41 |
asciilifeform |
shinohai: iirc user list nao only shows 'warm' peers |
13:42 |
asciilifeform |
aok |
13:42 |
shinohai |
Yeah but despite you being warm yer nowhere in my peer list. |
13:43 |
shinohai |
neither you, nor crtdaydreams |
13:43 |
asciilifeform |
shinohai: seeing you as hearsay atm so apparently not warm |
13:43 |
asciilifeform |
ditto crtdaydreams |
13:44 |
shinohai |
I suspect "yet another db problem" is at fault here, but haven't had time to investigate. On TODO list is make a bash script to export full WOT from blatta |
13:44 |
asciilifeform |
would be pretty handy. ditto for import. |
13:46 |
* |
shinohai apologizes for his crude thinking - guess you could also do with native python, but I'm just better @ bash things ... |
13:47 |
shinohai |
Add to above list billymg, also does not show in peer list |
13:48 |
asciilifeform |
shinohai: shows in asciilifeform's; see if actually connects when you have the latest blatta, w/ addrcast |
13:50 |
asciilifeform |
theoretically if all 3 have that one, and keys for ea. other, oughta connect |
13:51 |
asciilifeform |
(is whole point of addrcast) |
| |
~ 1 hours 19 minutes ~ |
15:10 |
signpost |
asciilifeform: bet ya have multiple shinohai keys as I had for various. |
| |
~ 34 minutes ~ |
15:44 |
awt |
asciilifeform: as signpost suggests, possibly you have multiple keys for some peers. |
| |
↖ |
15:45 |
signpost |
hilarious storm of join/part on my end atm |
15:45 |
* |
signpost checks logs |
15:46 |
signpost |
join/parts seem to have stopped. just a bunch of ADDRESS CASTs going on in logs. |
15:46 |
signpost |
maybe interaction between nodes of different version? |
15:48 |
* |
awt not seeing any |
15:53 |
signpost |
nah still churning |
15:54 |
signpost |
curious if the thing's preoccupied with these address cast messages and that's introducing delay in processing other messages. |
15:57 |
awt |
asciilifeform: noted, re: per cold peer |
15:59 |
awt |
signpost: I notice your ac messages come in pretty slowly, i.e. noticeable gap between when each arrives |
16:01 |
signpost |
cpu usage of the blatta process has approx doubled since update |
16:01 |
awt |
signpost: you can decrease the frequency by increasing the value of the address_cast_interval_seconds knob |
16:03 |
signpost |
will give that a try |
16:04 |
signpost |
eventually the hot loop in Server.start is probably going to need attention |
16:04 |
awt |
indeed |
16:04 |
signpost |
I wager I'm running this on one of the smaller boxes present |
16:05 |
signpost |
anyway just reporting what I see, and updating node quickly after release to help shake things loose. |
16:06 |
shinohai |
lol I'm running pest on a 32 bit chinsesium phone. |
16:07 |
awt |
signpost: looks like we have an infinite prod loop going |
16:09 |
signpost |
boog in lib/server.py#178, needs cast to int on the get_knob call |
16:09 |
signpost |
I raised the address cast interval to 600 |
16:09 |
signpost |
now seeing a flurry of "Peer with duplicate address or handle" messages |
16:11 |
signpost |
hm |
16:12 |
asciilifeform |
http://logs.bitdash.io/pest/2022-09-12#1012729 << why would this be an eggog condition tho? >1 key for a peer is 100% normal per spec |
16:12 |
bitbot |
Logged on 2022-09-12 15:44:48 awt: asciilifeform: as signpost suggests, possibly you have multiple keys for some peers. |
16:12 |
awt |
asciilifeform: explained upstack that get_keyed_peers func needs more work |
| |
↖ |
16:12 |
asciilifeform |
a |
16:12 |
signpost |
I now see my own IP in AT for phf and billymg. |
16:13 |
asciilifeform |
signpost: fwiw asciilifeform still sees at entry from 17 july for phf |
16:13 |
signpost |
asciilifeform: yup, I'm just reporting that it caused a bug in blatta. |
16:13 |
asciilifeform |
( for crtdaydreams, shinohai, PeterL -- atm asciilifeform sees at entry from 31 july ) |
16:14 |
asciilifeform |
( i.e. from right prior to asciilifeform crating up his station for move ) |
16:15 |
signpost |
deleting peerings with phf and billymg (neither of which were operative before I updated) stopped the foom of warnings, but the fact that their AT entries got updated with my station's own IP says something's broken. |
16:15 |
asciilifeform |
for bitbot -- 3 july. errybody else in asciilifeform's wot -- current |
16:15 |
asciilifeform |
lol defo bug |
16:15 |
signpost |
cpu usage now bouncing between say 25-50% of one core on the apu2 |
16:18 |
awt |
Sounds like potentially a dedup issue |
16:19 |
awt |
My at has an update from phf sept. 7th |
16:20 |
awt |
phf may have been connecting only via my station for some time |
16:21 |
asciilifeform |
awt: iirc he has own pestron and not sure whether supports addrcast etc |
16:22 |
signpost |
yep pretty sure he said as much |
16:22 |
awt |
yes. his pestron only broadcasts to a single peer, afact, not to all. |
16:23 |
asciilifeform |
a |
16:30 |
awt |
Just saw CPU spike to 99% |
16:31 |
awt |
meltdown |
16:32 |
signpost |
lol |
16:33 |
* |
signpost "nices" his blatta to 19 for the time being |
16:33 |
awt |
I suggest others not update for now |
16:34 |
signpost |
gonna leave mine up as is to see what other interesting stuff shakes loose |
| |
~ 1 hours 32 minutes ~ |
18:07 |
crtdaydreams |
asciilifeform: last seen 1st august |
18:07 |
crtdaydreams |
shinohai been broken since 17th july |
18:10 |
crtdaydreams |
err wot |
18:11 |
crtdaydreams |
address for peer awt set to ____ |
18:11 |
crtdaydreams |
oh dear god |
18:12 |
crtdaydreams |
asciilifeforms and thimbronions ips are changing almost every minute |
18:13 |
crtdaydreams |
notsure if result of as discussed above, but certainly not a feature |
18:14 |
crtdaydreams |
getting a lot of traffic from jonsykkel also |
18:14 |
crtdaydreams |
all AC |
| |
~ 17 minutes ~ |
18:32 |
* |
asciilifeform noticed that blatta nao takes ~10sec to respond to at cmd |
18:32 |
* |
asciilifeform will prolly end up rolling back to prev patch |
| |
~ 30 minutes ~ |
19:02 |
awt |
my at quite snappy still |
19:03 |
awt |
crtdaydreams: not entirely sure how ac is implemented in smalpest |
19:08 |
jonsykkel |
uses old format id gues its intepreting part of nonce as adres |
19:08 |
jonsykkel |
wil put a new version |
| |
~ 50 minutes ~ |
19:58 |
signpost |
backed off to 9973 for now. |
19:59 |
signpost |
!!help |
19:59 |
deedbot |
http://deedbot.org/help.html |
20:00 |
signpost |
deedbot seems to be doing fine on 9972 so I'll leave that station as is. |
| |
~ 3 hours 27 minutes ~ |
23:27 |
crtdaydreams |
$ticker btc usd |