We have an odd issue with Kratos, when performing the recovery flow, after clicking the recovery link in the email, we enter the password as prompted, and hit submit, the POST request then hangs until it times out (from our LB). The password does actually save though. Enabling debug on Kratos doesn't give much insight, no specific errors besides "context cancelled" from the timeout. End of the stacktrace is as follows;

time=2022-06-16T17:59:02Z level=info msg=Encountered self-service settings error. audience=audit error=map[message:context canceled stack_trace:
<http://github.com/ory/x/sqlcon.HandleError|github.com/ory/x/sqlcon.HandleError>
	/go/pkg/mod/github.com/ory/x@v0.0.358/sqlcon/error.go:93
<http://github.com/ory/kratos/persistence/sql.(*Persister).GetIdentityConfidential|github.com/ory/kratos/persistence/sql.(*Persister).GetIdentityConfidential>
	/project/persistence/sql/persister_identity.go:381
<http://github.com/ory/kratos/identity.(*Manager).Update|github.com/ory/kratos/identity.(*Manager).Update>
	/project/identity/manager.go:98
<http://github.com/ory/kratos/selfservice/flow/settings.(*HookExecutor).PostSettingsHook|github.com/ory/kratos/selfservice/flow/settings.(*HookExecutor).PostSettingsHook>
	/project/selfservice/flow/settings/hook.go:151
<http://github.com/ory/kratos/selfservice/flow/settings.(*Handler).submitSettingsFlow|github.com/ory/kratos/selfservice/flow/settings.(*Handler).submitSettingsFlow>
	/project/selfservice/flow/settings/handler.go:550
<http://github.com/ory/kratos/session.(*Handler).IsAuthenticated.func1|github.com/ory/kratos/session.(*Handler).IsAuthenticated.func1>
	/project/session/handler.go:458
<http://github.com/ory/kratos/x.NoCacheHandle.func1|github.com/ory/kratos/x.NoCacheHandle.func1>
	/project/x/nocache.go:18
<http://github.com/ory/kratos/x.NoCacheHandle.func1|github.com/ory/kratos/x.NoCacheHandle.func1>
	/project/x/nocache.go:18
<http://github.com/julienschmidt/httprouter.(*Router).ServeHTTP|github.com/julienschmidt/httprouter.(*Router).ServeHTTP>

Any suggestions?

this is caused by http timeouts

Thanks @high-optician-2097 Yep, which is probably our LB, as it has a request timeout of 60s to avoid long running requests, but I shouldn't expect the POST on setting a password to take longer than 60s

Depends on your hashing throughput probably

1 😅 It's just in our k8 Dev env, so traffic is usually just one or two requests. That said, you've given me an idea, and that if it's indeed hashing, we could be CPU throttling as we tend to keep the Dev environments pretty light on resources. I'll check that out first