[5930:Dispatcher]: [ERROR] Unhandled exception reached the top of main, device shutting down. Details follow:
[5930:Dispatcher]: Throw location unknown (consider using BOOST_THROW_EXCEPTION)
[5930:Dispatcher]: Dynamic exception type: boost::wrapexcept<boost::program_options::ambiguous_option>
[5930:Dispatcher]: std::exception::what: option '--config' is ambiguous and matches '--config-key', and '--configuration'
in the Dispatcher, QC-CHECK-RUNNER-QcCheck and QC-TASK-RUNNER-QcTask devices.
Simply changeing --config to either --config-key or --configuration did lead to
2020-05-20 11:25:10.529016 QC infologger initialized
2020-05-20 11:25:10.529027 Using config file ''
2020-05-20 11:25:10.529047 Creating a standalone QC topology.
[ERROR] error while setting up workflow: Ill-formed URI
[INFO] Process 6010 is exiting.
I guess simply the command I need to use changed a bit.
Can someone please give me some insight?
Hi
I think this is the same problem as reported here https://github.com/AliceO2Group/AliceO2/pull/3605 by Ruben.
Long story short, there was a PR adding a --configuration parameter to any DPL workflow, which collides with some other parameters due to the way boost handles those. I know that @eulisse is already aware of the problem, but seeing that it practically destroyed QC, I would vote for increasing priority on this.
Out of curiosity I checked if, as Giulio suggested in https://github.com/AliceO2Group/AliceO2/pull/3605, renaming configuration to dpl-configuration helps. And as far as I can judge it does. But I guess getting rid of allow_guessing goes more to the root of the problem.
@eulisse@tklemenz, let me mention that I am getting the same kind of problem with the current O2 dev and QC master branches. I would also vote for rising the priority of this issue, since I am practically unable to continue the QC development as we need an up-to-date O2 dev branch for the MCH-specific parts…
I am available to do testing if this can help you debugging the problem.
It is at the highest priority. Notice you need to use --config-key (in full now). I am surprised it does not work, did recompile everything with alibuild?
Latest alidist/master, O2/dev, and QC/master rebuilt with alibuild still gives
2020-05-23 10:21:47.135997 QC infologger initialized
2020-05-23 10:21:47.136009 Using config file ''
2020-05-23 10:21:47.136027 Creating a standalone QC topology.
[ERROR] error while setting up workflow: Ill-formed URI
[INFO] Process 14110 is exiting.
So, the problem is that FairMQ does not allow greedy parsing of options so everything worked so far basically by accident since --config from QC was prepended to --config-key from FairMQ… I will enable back the old behaviour and rename --configuration to --dpl-config,--cfg.