[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #1916 [Metrics]: Visualize torperf tries that time out
#1916: Visualize torperf tries that time out
---------------------+------------------------------------------------------
Reporter: arma | Owner: karsten
Type: defect | Status: new
Priority: normal | Milestone:
Component: Metrics | Version:
Keywords: | Parent:
---------------------+------------------------------------------------------
Comment(by karsten):
I graphed timeouts and failures in the attached torperf-failure-
timeout-2010-09-08.png . A failure is when a request finishes, but we
receive fewer bytes than we asked for. A timeout happens when the timeout
tool aborts a request in order to start the next one.
There were quite a few timeouts (up to 10%) in May 2010 when the Tor
network almost broke. But since then, the rate of failed or timed out
requests is at roughly 1% for 50 KiB requests. The failure and timeout
rates also decreased in the past few months as compared to last year. For
1 MiB and 5 MiB, the percentages of failures are higher, because a single
failed request means that 2.1% or 4.2% of all requests on that day failed.
In the past few months, there were at most 1 or maybe 2 failures per day.
Adding more graphs or changing existing graphs on the metrics website is
difficult right now, because Kevin and I are merging his GSoC stuff. But I
attached the R sources to plot this graph. We can easily make a new graph
in a matter of minutes if we think the failure or timeout rate might have
changed.
Is there anything in particular that I should look closer at?
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/1916#comment:1>
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