Opened 3 years ago

Closed 3 years ago

#20723 closed defect (wontfix)

Fix torperf file download from moria

Reported by: karsten Owned by: metrics-team
Priority: High Milestone:
Component: Metrics/CollecTor Version:
Severity: Normal Keywords:
Cc: Actual Points:
Parent ID: Points:
Reviewer: Sponsor:


A few days ago I started finding logs like the following written by CollecTor's torperf module:

2016-11-16 06:01:00,002 INFO o.t.c.c.CollecTorMain:66 Starting torperf module of CollecTor.
2016-11-16 06:04:08,431 WARN o.t.c.t.TorperfDownloader:260 The last timestamp line in '/srv/' is not contained in the new 
file downloaded from ''.  Cannot append new lines without possibly leaving a gap.  Skipping.

I figured it might be related to that server automatically redirecting http to https and CollecTor not following that redirect, so I changed to the https URL in

Now I receive the following log statements:

2016-11-17 18:03:03,215 WARN o.t.c.t.TorperfDownloader:267 Failed downloading and/or merging ''. Remote host closed connection during handshake
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        at org.torproject.collector.torperf.TorperfDownloader.downloadAndAppendFile( [collector-1.1.0.jar:1.1.0-d9e32d5]
        at org.torproject.collector.torperf.TorperfDownloader.downloadAndMergeFiles( [collector-1.1.0.jar:1.1.0-d9e32d5]
        at org.torproject.collector.torperf.TorperfDownloader.startProcessing( [collector-1.1.0.jar:1.1.0-d9e32d5]
        at [collector-1.1.0.jar:1.1.0-d9e32d5]
        at java.util.concurrent.Executors$ [na:1.7.0_111]
        at java.util.concurrent.FutureTask.runAndReset( [na:1.7.0_111]
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301( [na:1.7.0_111]
        at java.util.concurrent.ScheduledThreadPoolExecutor$ [na:1.7.0_111]
        at java.util.concurrent.ThreadPoolExecutor.runWorker( [na:1.7.0_111]
        at java.util.concurrent.ThreadPoolExecutor$ [na:1.7.0_111]
        at [na:1.7.0_111]
Caused by: SSL peer shut down incorrectly
        at ~[na:1.7.0_111]
        at ~[na:1.7.0_111]
        ... 17 common frames omitted

It seems that wget works just fine on that host to download that file.

Any idea what I could try?

I'll have to fix this in the next couple of days, or moria will delete older data and CollecTor won't be able to archive that. Therefore setting priority to high.

Child Tickets

Change History (3)

comment:1 Changed 3 years ago by iwakeh

Maybe, this helps: #20514.

comment:2 Changed 3 years ago by karsten

Aha, I'll try that tomorrow and report back. Thanks! :)

comment:3 Changed 3 years ago by karsten

Resolution: wontfix
Status: newclosed

I tried out various combinations of switching to the code suggested in #20514, passing,handshake and some other parameter to select the TLS version, switching to Java 8, and testing on different OSes. Some variants work, some don't. Given that we're going to switch to OnionPerf soon, we shouldn't spend more hours on debugging this. I turned off moria's Torperf instance, but we still have 2 instances left, one run by ln5 and run by me on a Tor VM. Closing as wontfix.

Note: See TracTickets for help on using tickets.