Skip to content

Problems with use of big timeout (or no timeout at all) in trace service #7147

@AlexPeshkoff

Description

@AlexPeshkoff

Was originally reported by Pavel Zotov privately.

Bug was found when working with new python testing tool. That tool has no timeout when using trace service.
When working with CS python tool hangs on exit to OS and a number of instances of CS remain running on the server.

It was found that all processes hang in same place - they wait indefinitely for more trace data, and there is no way to stop them. Bug has gone unnoticed when using fbsvcmgr / fbtracemgr cause both have 1 sec timeout. That <1 sec delay when closing utility appears to be not a problem for interactive user and was successfully ignored by all test suites.

It's trivial to reproduce without python slightly modifying out own utility. It was noticed that client also has serious problem - there is no way to interrupt Service::query(). fb_shutdown() also does not help.

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions