I'm not sure it can help with the problem, but starting with Oracle 11, OCI installs some custom signal handlers (even catching segmentation faults). They can be deactivated by setting options in the sqlnet.ora file. Here, we run all our OCI clients with:
DIAG_SIGHANDLER_ENABLED=FALSE DIAG_ADR_ENABLED=FALSE DIAG_DDE_ENABLED=FALSE Regards, Didier. On Wednesday, April 26, 2017 at 10:25:09 PM UTC+2, Tamás Gulácsi wrote: > > 2017. április 26., szerda 20:33:13 UTC+2 időpontban Pierre Curto a > következőt írta: >> >> Hello, >> >> I am unsure where to log the following issue: >> whenever I send an interrupt signal to a go process running an Oracle >> query, it panics with either "fatal: morestack on g0" or "fatal error: >> runtime: stack split at bad time". >> >> A sample code reproducing the issue is here >> <https://play.golang.org/p/f-XBPwNlqG>. Of course an Oracle instance and >> libs are required for it to work. >> >> Should it be reported to the maintainers of the Oracle package or the Go >> team? >> It seems to me it should be the later, but I want to make sure. >> > > You can report it at rana/ora, but I'll ask it here anyway: is this > avoidable? > My first instinct says no, we don't know what interrupt handlers are in > use in the OCI library - maybe re-install/shield some after instantiating > the library? > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
