[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #20440 [Metrics/metrics-lib]: endless wait in BlockingIteratorImpl
#20440: endless wait in BlockingIteratorImpl
---------------------------------+-----------------------------------
Reporter: iwakeh | Owner: iwakeh
Type: defect | Status: needs_review
Priority: High | Milestone: metrics-lib 1.6.0
Component: Metrics/metrics-lib | Version:
Severity: Normal | Resolution:
Keywords: | Actual Points:
Parent ID: | Points:
Reviewer: | Sponsor:
---------------------------------+-----------------------------------
Comment (by iwakeh):
Replying to [comment:13 karsten]:
> The fix looks good to me, even unrelated to this particular error case.
What if an application starts reading descriptors and decides at some
point to stop processing descriptors, possibly because a database
connection has broke or the disk has run full or whatever. In such a case
the application shouldn't be forced to request all descriptors from the
iterator until it's empty. And the application shouldn't be forced to
System.exit() just to kill the library thread. In a way, we're only using
a thread to read and parse descriptors to speed up the process and not do
all this when the application asks for another descriptor.
>
> I'm still curious though: how did you reproduce this bug? Can I have
the data you mention above that would reproduce it? (Or maybe I already
have it, but where?) Thanks!
The data is not public, but check your mail (Oct 24, 2016)
I think, it might be a concurrency issue or at least that affects metrics-
lib in addition.
For this discussion see ticket #21365.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/20440#comment:14>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs