zzz
eyedeekay, eche|off eche|on there's a major problem with the news being served at tc73n4kivdroccekirco7rhgxdg5f3cjvbaapabupeyzrqwv5guq.b32.i2p/news.su3
zzz
<i2p:release date="2023-01-10" minVersion="0.9.9" minJavaVersion="1.8">
zzz
<i2p:version>2.3.0</i2p:version>
zzz
<i2p:update type="su3">
zzz
<i2p:torrent href="magnet:?xt=urn:btih:f7d9ddeddb17ae728900c84e49debf7de5ebfa8a&dn=I2P+Update+2.2.1&tr=http://tracker2.postman.i2p/announc
zzz
e.php"/>
zzz
<i2p:url href="http://mgp6yzdxeoqds3wucnbhfrdgpjjyqbiqjdwcfezpul3or7bzm4ga.b32.i2p/2.2.1/i2pupdate.su3"/>
zzz
</i2p:update>
zzz
</i2p:release>
zzz
the version is 2.3.0 but the magnet and direct downloads are for 2.2.1
zzz
the update fails "version mismatch"
eyedeekay
Ack I'll figure out what's happened and fix it tonight
zzz
thanks. whats in i2p.newsxml looks right (except for the release date)
zzz
commit 1829ed012e0f7d8cde4b3a44cfbdf2c65caef7e2
zzz
Author: eyedeekay <idk@mulder>
zzz
Date: Tue Jul 11 17:27:23 2023
zzz
fix version numbers and blog post URLS
zzz
it looks like eche|on rebuilt on 7/11 but without your fix?
zzz
and how did you guys go 11 days without realizing the error after the release, and then eche still didn't fix the news right and nobody caught it?
zzz
and 3 months later still busted?
zzz
I don't know how or why you got as much of the net on 2.3.0 as you do
eyedeekay
Easy-install defaults to my news but otherwise it's tc7, must have fallen to the backup at some early point
RN
building -8 Σ:Đ
RN
2.3.0-8 available on I2Peek-a-Boo.i2p
RN
what's the status on Macintosh installer and or updating. Got someone stuck on 1.9.0
zzz
eche|off, eche|on, please let me know when you fix it so I may retest
zzz
or eyedeekay if you fix it let me know
eche|off
what fix?
eche|off
currently I get no email, don't know why
zzz
see 12 hours above ^^^
zzz
news broken at 2.3.0 release late June
zzz
eyedeekay fixed it in newsxml July 11
zzz
you rebuilt news July 11 but _without_ the fix
zzz
metadata has 2.3.0 release number but 2.2.1 magnet and URL
eche|off
oh, I di dnothign on that for ages now
eche|off
no time for anything
zzz
well, it's wrong and been wrong for 4 months, and people can't update to 2.3.0 because of it
zzz
please pull newsxml
eche|off
as soon as I got time
zzz
verify in releases.json that the magnet and URL are right
eche|off
tomorrow or weekend
zzz
rebuild news
zzz
let me know
zzz
thank you
zzz
we're failing at the basics which is getting releases to people
eche|off
I do not know howtro verify. if the pull pulls worng data
eche|off
but we will see later on
zzz
well, I wasn't here, but apparently he fixed it on July 11, but you didn't get the fix
zzz
here is how to verify so you know:
zzz
magnet starts with b3a8
zzz
url has 2.3.0 in it
zzz
in file data/releases.json
zzz
subsession LS MR is up for review
zzz
and working on the next one which will be about 700 lines of reversions
zzz
as for the news fix, it's been broken for 4 months, so what's another couple days...
zzz
eyedeekay, next big 'design intent' question
zzz
I was going to wait until we finished the multihome discussion but I'm stuck on it now
zzz
can subdbs have RIs in them or not?
zzz
whichever the answer is there's a lot of inconsistencies in the code to be resolved
zzz
this is another big topic right up there with multihome
dr|z3d
possible issue with the latest patchset:
dr|z3d
Error processing job [Lookup LeaseSet for Client] on thread 8
dr|z3d
java.lang.NullPointerException: Cannot invoke "net.i2p.router.networkdb.kademlia.FloodfillNetworkDatabaseFacade.lookupDestination(net.i2p.data.Hash, net.i2p.router.Job, long, net.i2p.data.Hash)" because the return value of "net.i2p.router.client.ClientConnectionRunner.getFloodfillNetworkDatabaseFacade()" is null
dr|z3d
at net.i2p.router.client.LookupDestJob.runJob(LookupDestJob.java:189)
dr|z3d
at net.i2p.router.JobQueueRunner.runCurrentJob(JobQueueRunner.java:118)
dr|z3d
at net.i2p.router.JobQueueRunner.run(JobQueueRunner.java:67)
zzz
hmm
dr|z3d
this line:
dr|z3d
// TODO tell router this is an encrypted lookup, skip 38 or earlier ffs?
dr|z3d
_runner.getFloodfillNetworkDatabaseFacade().lookupDestination(_hash, done, timeout, _fromLocalDest);
zzz
reproduced here, thanks, will work on it
zzz
missed a whole bunch of spots that need null checks after that MR
dr|z3d
ok, good, not a heisenbug then.
zzz
no, thankfully
zzz
oh good, the fixes are mostly revert revert revert
eyedeekay
Sorry you've been talking to my bouncer all morning, outage, no connection until now, signing and uploading thew news now
zzz
dr|z3d, fixed/briefly tested/pushed on top of MR 136, thanks again
dr|z3d
nice, zzz, thanks.
RN
some interesting wrapper log from 2.3.0-08 cake.i2p/view/s0ySHANnrh_MLHL36n6MybVq4F4fwdSDBfYSygpqV_2Ep1PCrxBw/s0ySHANnrh.txt
eyedeekay
news should be fixed up on tcy finally
eyedeekay
Hmmmm... BSD?
RN
Ubuntu 22.04.3 LTS (GNU/Linux 5.15.0-1043-realtime x86_64)
RN
not my box, but I have shared access
RN
it is moving ~250KB/s+ up/down so performing well. curious if we should apply the suggested fixes from the log entry
RN
revision: e204c50....
eyedeekay
You can, IIRC we use jbigi way less than before but still use it in a few important places, zzz would know better than me on that one
eyedeekay
Hmm, that ELFCLASS32 thing seems like a possible cause...
eyedeekay
Does `file /tmp/i2p-CaaN0SAK.tmp/libjcpuid.so` agree that it's a 32-bit ELF?
eyedeekay
If that's the case then the solutions aren't likely to work
eyedeekay
Should say something like: libjcpuid.so: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked
dr|z3d
eyedeekay: you want to put teddit.i2p on the canon homepage?
dr|z3d
if we're trying to migrate people off reddit, probably a good move.
RN
hrrrrm... 'no such file or directory'
RN
that's not great.
eyedeekay
dr|zed yeah might be a good idea
dr|z3d
check /tmp/ for random i2p- dirs, RN.
eyedeekay
or maybe `find /tmp -name libjcpuid.so -exec file {} \; 2>/dev/null` or possibly also `sudo -u i2psvc find /var/lib/i2p -name 'libjcpuid*.so' -exec file {} \; 2>/dev/null`
eyedeekay
thanks
RN
yeah, got the dir /tmp/i2p-CaaN0SAK.tmp/ with several jetty dirs and a snark dir, but no libjcpuid.so in there
eyedeekay
Well that would also cause that error at least according to StackOverflow
eyedeekay
SO link with similar message: stackoverflow.com/questions/24355833/openjdk-64-bit-server-vm-warning-you-have-loaded-library-which-might-have-disab
eyedeekay
zzz been having some spam filter issues finally found your reply re: multihomeDb and disableSubDbs and replied back
zzz
RN et al the real problem is the first one - the stack guard thing
zzz
the 2nd error was just it trying the 32 bit lib after the 64 bit one failed
zzz
please file a ticket with the logs from the paste and full version info from /logs in console