You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
gRPC _Rendezvous w/ 'ABORTED' status propagated from 'Session.run_in_transaction' #3562
The code for Session.run_in_transaction expects that the ABORTED bit will be propagated as an instance of google.gax.errors.GaxError: somehow, the underlying GAPIC code is raising a gRCP _Rendezvous instead.
changed the title [-]Unexpected 'ABORTED' error propagated from 'Session.run_in_transaction'[/-][+]gRPC _Rendezvous w/ 'ABORTED' status propagated from 'Session.run_in_transaction'[/+]on Jun 28, 2017
Hi this is very annoying, to have a whole bunch of exceptions thrown in my application whenever a transaction is aborted. Is there any workaround to this issue at the moment?
So, the issue is that the streaming_read and execute_streaming_sql calls are propagating the _Rendezvous error, rather than wrapping them up in a GaxError as the commit call does.
Repeating this back to ensure I understand: Basically this is a situation where one call wraps the errors while others do not? Is this a configuration difference? Something else?
To further clarify: #3738 is sufficient to solve this specific case because
spanner has significant manual code above the gapic, so the mapping can be
done there. google.api.core will provide a iterator wrapper for
unary-streaming that can be used by gapic.
Fixing this issue that's blocking Python Spanner Beta has the greatest urgency, so it's good to hear that #3738 is sufficient to fix it.
That said, we'll need to fix this for GAPIC-only libraries. @jonparrott can you show @eoogbe what needs to be done to port your work to GAX, so that she's also familiar with the fix?
Activity
[-]Flaky Spanner system test[/-][+]Unexpected 'ABORTED' error propagated from 'Session.run_in_transaction'[/+]tseaver commentedon Jun 28, 2017
The code for
Session.run_in_transactionexpects that theABORTEDbit will be propagated as an instance ofgoogle.gax.errors.GaxError: somehow, the underlying GAPIC code is raising a gRCP_Rendezvousinstead.[-]Unexpected 'ABORTED' error propagated from 'Session.run_in_transaction'[/-][+]gRPC _Rendezvous w/ 'ABORTED' status propagated from 'Session.run_in_transaction'[/+]tseaver commentedon Jun 28, 2017
@lukesneeringer Do we need to be regenerating the GAPIC code?
volokluev commentedon Jul 20, 2017
Hi this is very annoying, to have a whole bunch of exceptions thrown in my application whenever a transaction is aborted. Is there any workaround to this issue at the moment?
tseaver commentedon Jul 25, 2017
@vladkluev I'm hopeful we will have a fix pushed out shortly (this bug blocks PR #3663).
tseaver commentedon Jul 25, 2017
So, the issue is that the
streaming_readandexecute_streaming_sqlcalls are propagating the_Rendezvouserror, rather than wrapping them up in aGaxErroras thecommitcall does.bjwatson commentedon Jul 25, 2017
@tseaver I'm discussing this with @lukesneeringer offline.
lukesneeringer commentedon Jul 25, 2017
Repeating this back to ensure I understand: Basically this is a situation where one call wraps the errors while others do not? Is this a configuration difference? Something else?
17 remaining items
lukesneeringer commentedon Aug 8, 2017
google.api.core.theacodes commentedon Aug 8, 2017
bjwatson commentedon Aug 8, 2017
Fixing this issue that's blocking Python Spanner Beta has the greatest urgency, so it's good to hear that #3738 is sufficient to fix it.
That said, we'll need to fix this for GAPIC-only libraries. @jonparrott can you show @eoogbe what needs to be done to port your work to GAX, so that she's also familiar with the fix?
theacodes commentedon Aug 8, 2017
Following up internally.
bjwatson commentedon Aug 8, 2017
Thanks for the update @jonparrott. I've responded internally.
What needs to be done to close this specific issue for Spanner? Will you do it or should @tseaver?
theacodes commentedon Aug 8, 2017
It's up to @tseaver (see here) but i'm happy to help.
bjwatson commentedon Aug 8, 2017
Thanks, Jon.
Workaround for #3562.
tseaver commentedon Aug 9, 2017
I will be merging the workaround in #3663 for spanner tonight.
bjwatson commentedon Aug 9, 2017
Thanks @tseaver!
tseaver commentedon Aug 10, 2017
Via #3663.