Tribler stopped on error code#
Tribler stopped on error download#
Tribler stopped on error mac#

This in turn causes the circuits_needed to go to 0 when a download is removed. If there are preexisting tunnels (like a download already going) build_tunnels() will be skipped, num_hops_by_downloads won't be incremented and is no longer accurate. The can_create_engine_wrapper() first checks if there is a need to build tunnels, and then calls build_tunnels(). Which in turn gets called only once, from deep within can_create_engine_wrapper(). It gets reduced when a download is removed. So how does the num_hops_by_downloads counter work? It should count the number of active downloads for a given number of hops. So that coincides with the on_download_removed() code. And the problem occured some 10 minutes after a completed download was removed and the last tunnels have expired.

So I add downloads by bash scripts & curl, and I have a systemd timer that runs a check for completed downloads and removes them. But it is only made 0 when num_hops_by_downloads goes to 0.

Tribler stopped on error code#

There is only one place in the code where this assignment is made triblertunnel/community.py:on_download_removed(). I added a debug statement to do_circuits. I'm going to add some debug log lines to debug further. The version I used here is recent but not bleeding edge, so please someone tell me if something like this was recently fixed. I expect this is somewhere in the create_engine_wrapper or that huge tunnel community callback that monitors download. After 10 mins this new tunnel is closed due to age and there are 0 tunnels and multiple active downloads again. If I add another download, the tunnel community is triggered to build some more tunnels and downloading resumes on all downloads. Indeed there are zero tunnels, no tunnels are being built, but there are active downloads. And TriblerTunnelCommunity keeps telling us it wants zero tunnels. Notice that in the beginning there are circuits to send the DHT requests, but after some tunnels get closed, there is no longer a tunnel to send the request on. I managed to get this log segment of around the time that it happend. I however my downloads are always stuck at 0 and don't recover. I can semi-reliably observe a crash to zero after some 1h to 1h45min after starting tribler. I run on a virtual server (with no USB drives attached), I've checked that I don't run into I/O limits. I experience a drop to zero also, but I expect that in my situation it is actually a different problem.

Do we need something complex like ionice or trickle+ NBD?

Tribler then closes responsively within Please try to make a reproducible slow IO testing ecosystems for this within Jenkins/nosetest.

wait longer, download resumes, finishes.

wait for 30 seconds to see the other download jump to 1 Kbps, then crash to 0 again.

watch as both downloads experience a 'bandwidth crash'.

start second download to slow USB stick.

download on fast internal harddisk stable download.

Nope, it's merely 30minutes of backlog USB stick pending writes it seems. My understanding was that Tribler has gone in some error state, went corrupt, ran out of file descriptors, something.

Tribler stopped on error mac#

In the past I've seen that on Mac the shutdown can take over 30minutes. It is a Windows-compatible large format USB drive, mounted on Mac. If you understand when it fails, Tribler is amazing.Ĭonducted an experiment to understand if the USB bottleneck theory holds. I found that when the app quit I can still found the downloaded data in the cache (in '/Downloads/' + bundleIdentifier ) as a tmp file. I want, when the app force quit to be able to resume the download. I'm using TCBlopDownloadSwift for the download ( I converted it for swift 2.0 and it works fine) and I made the config for a background download. My app have to download a pretty big file ( 390Mb),