EOSERV Bug Tracker > Bug #146: SLN thread instability

Bug #146: SLN thread instability

SLN thread instability
ID #146
Submitter Sausage
Product EOSERV
Severity Normal
Status CLOSED, FIXED
Submitted 4th Oct 2012
Updated 6th Oct 2012
Sausage Submitter
Developer 12 years, 13 weeks ago

SLN thread seems to be very prone to failing and breaking itself or the entire server. Unsafe cross-thread memory accesses should be fixed to see if the problem resolves itself, as well as moving to a task pool model.

Not confirmed if the problem is caused by unsafe memory accesses, or external factors making the thread fail.

Comments

Sausage Submitter
Developer 12 years, 13 weeks ago

I'd say the most likely candidate is trying to re-register the timer event from inside of the thread.

Proposed fix is to move all unsafe code out of the thread by building the HTTP request before-hand and signaling the main thread when finished (SLN specific), doing the same and moving to a thread-pool based architecture (fun), or removing the thread and making an asynchronous HTTP request system that runs in the main thread.

Updated Status to CONFIRMED

Sausage Submitter
Developer 12 years, 13 weeks ago

Fixed in r353, hopefully.

Updated Status to CLOSED, FIXED

Add Comment

Please don't post unless you have something relevant to the bug to say.
Do not comment to say "thanks" or "fix this please".

Please log in to add comments. EOSERV Bug Tracker > Bug #146: SLN thread instability