[tahoe-lafs-trac-stream] [Tahoe-LAFS] #2972: Understanding the behavior of the Magic-Folder frontend by reading the logs is difficult or impossible

[Tahoe-LAFS]

#2972: Understanding the behavior of the Magic-Folder frontend by reading the logs
is difficult or impossible
---------------------+---------------------------
 Reporter:  exarkun  |          Owner:
     Type:  defect   |         Status:  new
 Priority:  normal   |      Milestone:  undecided
Component:  unknown  |        Version:  1.12.1
 Keywords:           |  Launchpad Bug:
---------------------+---------------------------
 As a developer, I want to be able to understand the decisions made by and
 the actions taken by the magic-folder frontend implementation.  I want to
 be able to debug failures in the test suite (unit tests and integration
 tests), both existing (eg when extending support to a new platform) and
 new (introduced by other development efforts).  I also want to be able to
 debug failures found in the wild - eg in a Tahoe-LAFS process running on a
 developer or even end-user machine.

 To this end, I want some kind of logs which clearly explain everything
 Magic-Folder is doing.

 The Magic-Folder implementation currently emits some logs.  However:

   * They are entirely ad hoc.  Information is logged if someone felt like
 logging it at some point.  This probably often corresponds to pieces of
 the implementation that some developer found difficult or that are near a
 previously fixed bug.  The log messages are largely prefixed with "Magic-
 Folder" but they are otherwise freeform text with no discernable
 conventions being followed.
   * They are only available from the Foolscap log which is often not
 available.  Extracting it requires specialized tools and is a time-
 sensitive operation (wait an hour and chances are you're not going to be
 able to get what you want).
   * Causal information between events is almost completely lost as the
 events are flattened into a linear sequence of text events.  There's no
 straightforward way (and sometimes no way at all) to connect two events
 and say that one caused the other.  Therefore, many events end up being
 completely context-free and therefore almost meaningless (eg - the
 uploader decided to upload a file?  great - but *why*?).

--
Ticket URL: <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2972>
Tahoe-LAFS <https://Tahoe-LAFS.org>
secure decentralized storage