#Why I don't see breadcrumbs for this exception?
75 messages · Page 1 of 1 (latest)
@drowsy tree it shouldn't be the CaptureExceptions. It could be that no breadcrumbs were generated until that point. The error itself tells me that it happened during startup, so I'm assuming that would be the case. I would tag @floral mango just in case to confirm, since he's working on the Unity SDK.
no it is impossible that breadcrumbs are not generated until that point
from what this looks to me you called CaptureMessage on this, right?
and breadcrumbs should be there anyway..strange
did you enable the new Logs feature?
the current code is CaptureException, but the live one could have been different maybe
yes it works in the other events
this was a surprise
which is more or less at the same time of the one above
if's a bit iffy, if you enable Logs then it automatically skips adding breadcrumbs. you then have to opt-in breadcrumbs on top of it. so we don't bloat and send stuff twice
currnetly t his is what I got
(btw structured logging is new to me I don't know what it is and I didn't c heck)
this is what I would expect based on the screenshot
yeah so the question remains 😄
the setup is identical for the 2 calls
however some time I use capture exception some time capture message
in fact I can never make up my mind about the two
if you have an exception object, i.e. try-catch, go with CaptureException
I oftern do this instead
SentrySdk.CaptureMessage(e.Message, (scope) =>
{
scope.SetTag("exception", e.Message);
scope.SetExtra("stackTrace", e.StackTrace);
});
but yeah the first one was a CaptureException
if you go back in time in the same event, some of them have breadcrumbs
and they are all from the same release
ha.. you add the stacktrace as an additional data sting
do you let the SDK self-initialize? or do you call Init yourself at some point?
ah!
there.. lol sneaky
are you ever clearing that scope?
we do stuff at the begin, but nothing changes during execution
breadcrumbs "exist" on the scope. so if you create a new scope or you clear the scope, the breadcrumbs disappear
alternatively, if you've implemented a SentryOptions.SetBeforeBreadcrumb filtering might kick in and just not let any through
it doesn't explain why not working is an exception
again that would be true for all the events
we use ConfigureScope but we don't create new scopes
interesting @floral mango in some event of the same kind there are breadcrumbs : https://core-loop-games-inc.sentry.io/issues/7153355937/events/3d772009c249466fa74f0b5311169f24/
yes. which leads me to believe that there is something being done to the scope or filtering at some point in your game.
the only think you made me think of is if the game reloads. So we have an option to full reload the game, returning to the same path where this is executed. However the code is deterministic, doens't change. Why would the same code not work the second time?
only explanation because otherwise literally this is done at the begin of the game, few lines after the sentry init
is it calling SentrySdk.Init again?
i don't think so, no
few lines after was the wrong way to put it, what I meant is that this is early game stuff
few lines after log wise hehe
you should be able to test this locally, right? start the game, logs something, capture an error/exeption, reload the game, log some more, capture something else. this should show up right away in sentry
Yes I can do that test
@floral mango Hijacking the thread for another questio: is it possible to let sentry recognise different "users"from the same process? let's say I want each thread of a process to be recognised as a different "user"
i'm not sure i follow. you want to set a different user depending on which thread the error comes from?
yes
special case for our game server, where a t hread = a match
basically right now we don'tt know exactly how many matches are affected because sentry counts the process, not the single match
@floral mango another issue (recycling thread). I am trying to understand why it's a couple of day suddenly UCB doesn't want to finish the iOS build. The project build correctly, xcode compiles but then I get an " error: API request failed". AI is firmtly convinced that is due to Sentry (I didn't even mention it) because somehow it's failing to upload the symbols
do you have access to the xcode build logs?
let me echeck
AI was right
Writing logs to './sentry-symbols-upload.log'
INFO 2026-03-28 17:20:56.519170 +00:00 Loaded file referenced by SENTRY_PROPERTIES (sentry.properties)
INFO 2026-03-28 17:20:56.519676 +00:00 sentry-cli was invoked with the following command line: "./sentry-cli-Darwin-universal" "debug-files" "upload" "--il2cpp-mapping" "--include-sources" "/Users/buildbot/Library/Developer/Xcode/DerivedData/Unity-iPhone-emiqvwlejwlpsqgscqygftoiinfu/Build/Intermediates.noindex/ArchiveIntermediates/Unity-iPhone/BuildProductsPath/Release-iphoneos"
error: API request failed
Caused by:
sentry reported an error: You do not have permission to perform this action. (http status: 403)
Command PhaseScriptExecution emitted errors but did not return a nonzero exit code to indicate failure```
never seen this before
I suppose is not reaching the auth key or something?
looks like the auth-key provided is missing the permissions to upload the symbols
I pass it through the sentry config file so I am not sure
nothing comes to my mind at glance, I will need to investigate because ONLY the ios is failing, all the other builds are not failing, but they are not on UCB either
also it started all of a sudden
to unblock yourself right now, in the config window there is a "ignore cli errors" that you can check so you at least can build locally
but they are not on UCB either
that's kinda relevant. if you do not provide the cli options as asset you need to make sure they are available in the UCB somehow. same as the option configuration i added a cli option configuration callback. so you can put your own code there to get the auth token from the environment (as you probably don't want to check it in and keep it secret)
but they are as an asset, what I meant is that the proejct is the same across platforms and building machines
this has always been like this
(just enabled ignore clie errors anwyay)
anyway for the time being the ignore cli errors is working
is this asset available on the UCB or is that on something like .gitignore
the project is identical across the platforms, it would be ignored everywhere
that's what i mean. then it will most likely fail everywhere in UCB