&eche|on
&zzz
+R4SAS
+RN
+RN_
+StormyCloud
+T3s|4
+acetone
+dr|z3d
+eche|off
+not_bob_afk
+orignal
+postman
+snex
+wodencafe
Arch
BubbRubb
Chrono
Danny
DeltaOreo
FreefallHeavens
Irc2PGuest16752
Irc2PGuest33667
Irc2PGuest51880
Onn4l7h
Over
Sisyphus
Sleepy_
Teeed
aargh2
ac9f_
b3t4f4c3__
b4dab00m
bak83_
duanin2
duck
john231
leopold
makoto
mareki2p_
nilbog-
poriori_
profetik1
r00tobo
rapidash
shiver_
solidx66
thetia
u5657
uop23ip
vivid_reader56
x74a6
xHarr
RN
didn't I tell you about something similar not that long ago dr?
dr|z3d
I don't recall, RN
zzz
dr|z3d, any tunnel shenanigans like multiple proxy tunnels? or you're stopping/starting manually?
dr|z3d
nothing unusual zzz. auto-close on idle.
dr|z3d
it's not a globally present issue, some routers seem fine, others won't update subs.
dr|z3d
not sure if the error posted above is related.
dr|z3d
but I suspect it might be.. seems to be related to the SVC_HTTP_PROXY variable.
zzz
you can see if it's registered on /debug port mapper section, HTTP. I just stopped and restarted proxy and saw it go away and come back on /debug
dr|z3d
seems fine there
dr|z3d
I've turned off close on idle on the troublesome router, still getting told HTTP proxy isn't running when I attempt to manually update subs.
dr|z3d
unlikely, but it's not related to git.idk.i2p/i2p-hackers/i2p.i2p/-/commit/a6ad5251973c39c976434d3c2a4304487163b81e ?
term99
one of my routers is having problems updating subs also, gives me an error of "HTTP proxy is not running!"
zzz
dr|z3d, where are you seeing the message?
dr|z3d
zzz: in susidns when attempting to manually update subs.
zzz
logs or UI?
dr|z3d
UI
dr|z3d
susidns/addressbook is pretty limited (non-existent) for logging.
zzz
I couldn't find where we have a UI msg for that
dr|z3d
SusbscriptionsBean.Java
dr|z3d
line ~150
dr|z3d
> if (content != null && content.length() > 2 && _context.portMapper().isRegistered(PortMapper.SVC_HTTP_PROXY)) {
zzz
ok so it just says 'subscriptions saved' if the proxy isn't up
zzz
right?
dr|z3d
no
dr|z3d
if (content != null && content.length() > 2 && _context.portMapper().isRegistered(PortMapper.SVC_HTTP_PROXY)) {
dr|z3d
_context.namingService().requestUpdate(null);
dr|z3d
message = _t("Attempting to update addressbook from subscription sources...");
dr|z3d
// + "<img height=\"1\" width=\"1\" alt=\"\" " +
dr|z3d
// "src=\"/addressbook/?wakeup=1&nonce=" + nonce + "\">";
dr|z3d
} else if (!_context.portMapper().isRegistered(PortMapper.SVC_HTTP_PROXY)) {
dr|z3d
message = _t("Cannot update subscriptions: HTTP proxy is not running!");
dr|z3d
}
zzz
that's your mod
zzz
anyway
zzz
I see something really fishy on one of my routers
dr|z3d
related to the above? :)
zzz
maybe. the addressbook/ config dir is in ~/i2p, not ~/.i2p
zzz
could you check yours?
dr|z3d
yup, same.
zzz
there's no way it should be in the base dir
dr|z3d
addressbook dir in ~/i2p/ with all the bits 'n bobs in there.
zzz
check and see if the location correlates with which routers have the issue
dr|z3d
definitely does.
dr|z3d
I can do an instant comparison and see that routers that don't have that folder located in ~/i2p don't have the problem.
dr|z3d
if that router you're seeing the behavior on happens to be a faily new install, I bet it hasn't pulled any addresses from any subs...
zzz
yeah the only one of mine with the issue is pretty new
dr|z3d
Right, let's have another look at your CLI stubs commit..
zzz
no
zzz
it's a context issue
zzz
susi is off in her own context
dr|z3d
hmmkay.. I was tending towards thinking it might be a buildfile issue, but ok.
zzz
maybe related to the tomcat version I updated to before the release, or maybe not
dr|z3d
oh, could be.
dr|z3d
there's something else you can check that may be related.
dr|z3d
see if your router with the issue displays identicons in the main listings.
dr|z3d
(in susidns)
zzz
I usually don't do updaterWithJetty so I'm not even sure which of my routers have what
dr|z3d
what I can also tell you is that the issue isn't limited to newly installed routers.. I've got a router that's been installed for a couple of years at least that recently started displaying the same issue.
zzz
you know, you might have a point, the CLI thing actually might cause classes to get loaded too early
zzz
will take a look
dr|z3d
I figured that's the only thing that's changed recently re addressbook.
zzz
on that older router, you have addressbook/ in both .i2p and i2p ?
dr|z3d
yup
zzz
ok
zzz
same router, good on java 8, bad on java 17
zzz
2022/06/17 10:10:15 | java.lang.Exception: I did it
zzz
2022/06/17 10:10:15 | at net.i2p.I2PAppContext.<init>(I2PAppContext.java:216)
zzz
2022/06/17 10:10:15 | at net.i2p.I2PAppContext.getGlobalContext(I2PAppContext.java:142)
zzz
2022/06/17 10:10:15 | at net.i2p.addressbook.DaemonThread.run(DaemonThread.java:60)
dr|z3d
ah, that's the other thing I thought about, java version issue. that correlates to what I'm seeing.
zzz
it's a classloader issue which is my least favorite thing
dr|z3d
joy!
SilicaRice
what happens if 2 routers publish different(ly signed) leasesets for the same offline key?
zzz
java 11 fails. older tomcat no change
dr|z3d
nasty bug, glad we caught it. could have left a bunch of users wondering why their subs weren't updating.
zzz
still not sure what changed or when
zzz
got it fixed, just doing final testing
dr|z3d
oh, nice. well done.
dr|z3d
could have been either jetty or tomcat updates I guess. maybe the latest jetty patches?
zzz
I went back to tomcat 54, no change
zzz
ok, try -7
dr|z3d
oh, -7 eh? good stuff.
dr|z3d
*pat pat*
zzz
still a little stumped. but susidns webapp has always been a little squirrely and different from the others
dr|z3d
maybe time to give it an overhaul
dr|z3d
hasn't really been touched in years, has it
dr|z3d
?
dr|z3d
damn that's an efficient fix.
dr|z3d
the search function could use inclusion of b32 in the search. and it's currently case sensitive, which it shouldn't be.
dr|z3d
"I wonder if this b32 correlates to anything in my addressbook?" kinda thing..
dr|z3d
patch confirmed working, thanks zzz.
dr|z3d
it looks like the issue was introduced around the end of may, judging by my susidns logs.
dr|z3d
last successful hosts fetch before the patch was on the 31st.
zzz
ok, thanks for testing and the initial report. CLI went in on the 26th. I didn't try backing that out yet
zzz
guess I need to to assess whether the bug was in the release
zzz
or I could just test the release, duh
zzz
lol
zzz
release is fine. The CLI was the cause, indeed
zzz
wtf
zzz
good news, anyway
term99
susidns sub. update fixed, ty for the updates
zzz
thanks for the report yesterday term99, always helps to have somebody say 'me too'
term99
but your welcome, glad to report