Closed adur1990 closed 6 years ago
Not dealing with "database busy". I think the right fix here is to pass the rhizome_bundle_status back to the caller and make them deal with it.
However, using the meshms restful API will avoid this problem completely.
We were sending using the restful api... only receiving nodes were periodically checking list conversations for status. Also the crashing node was only a relaying one.
We will try to port our periodic check scripts to restful even though no manual command should have such fatal effects imo :)
Recent improvements in the documentation and testing of RESTful APIs make it easier for scripts to use utilities like curl(1) instead of invoking the servald command line, which as @lakeman points out will not suffer from database lock errors.
The ideal option would be to provide a servalc (Serval Client) command-line utility that acted as a REST HTTP client of the daemon, so scripts could use it and get their results back formatted just like the existing CLI. However that is beyond the scope of this issue.
Closing.
rhizome_add_manifest_to_store() SIGABRT with commit 7a9b6d5d722713bdaf8f2a67cea63222d728d87e of asserts branch.
Traceback:
node3.log.zip