transmission-daemon: completion.c:330: tr_cpFileIsComplete: Assertion `tr_torBlockPiece( tor, lastBlock ) == file->lastPiece' failed.

[turbo]

This is SVN version 7789. But looking at completion.c, this bug SHOULD exist in 1.50b2 as well...

[turbo]

Backtrace:

(gdb) bt #0 0xb7f5e410 in ?? () #1 0xb798d248 in ?? () #2 0x00000022 in ?? () #3 0x08b92320 in ?? () #4 0xb7d2587b in write_nocancel () from /lib/tls/i686/cmov/libpthread.so.0 #5 0x0806b14b in tr_evbuffer_write (io=0x894c030, fd=171, howmuch=34) at peer-io.c:252 #6 0x0806c796 in tr_peerIoTryWrite (io=0x894c030, howmuch=1024) at peer-io.c:793 #7 0x0806c937 in tr_peerIoFlush (io=0x894c030, dir=TR_CLIENT_TO_PEER, limit=1024) at peer-io.c:823 #8 0x08063e06 in tr_bandwidthAllocate (b=0x80a9c80, dir=TR_CLIENT_TO_PEER, period_msec=500) at bandwidth.c:221 #9 0x08072303 in bandwidthPulse (vmgr=0x80aada8) at peer-mgr.c:2368 #10 0x0805fa8b in timerCallback (fd=-1, event=1, vtimer=0x80aadc0) at trevent.c:307 #11 0x0808eb2e in event_base_loop (base=0x80ab0e0, flags=0) at event.c:392 #12 0x0808ee2a in event_loop (flags=0) at event.c:468 #13 0x0808ee42 in event_dispatch () at event.c:406 #14 0x0805f889 in libeventThreadFunc (veh=0x80aacc0) at trevent.c:251 #15 0x08051497 in ThreadFunc? (_t=0x80aad50) at platform.c:106 #16 0xb7d20267 in start_thread () from /lib/tls/i686/cmov/libpthread.so.0 #17 0xb7cb460e in clone () from /lib/tls/i686/cmov/libc.so.6

[turbo]

Better backtrace?

(gdb) bt
#0  0xb7f5e410 in ?? ()
#1  0xb798d248 in ?? ()
#2  0x00000022 in ?? ()
#3  0x08b92320 in ?? ()
#4  0xb7d2587b in __write_nocancel () from /lib/tls/i686/cmov/libpthread.so.0
#5  0x0806b14b in tr_evbuffer_write (io=0x894c030, fd=171, howmuch=34) at peer-io.c:252
#6  0x0806c796 in tr_peerIoTryWrite (io=0x894c030, howmuch=1024) at peer-io.c:793
#7  0x0806c937 in tr_peerIoFlush (io=0x894c030, dir=TR_CLIENT_TO_PEER, limit=1024) at peer-io.c:823
#8  0x08063e06 in tr_bandwidthAllocate (b=0x80a9c80, dir=TR_CLIENT_TO_PEER, period_msec=500) at bandwidth.c:221
#9  0x08072303 in bandwidthPulse (vmgr=0x80aada8) at peer-mgr.c:2368
#10 0x0805fa8b in timerCallback (fd=-1, event=1, vtimer=0x80aadc0) at trevent.c:307
#11 0x0808eb2e in event_base_loop (base=0x80ab0e0, flags=0) at event.c:392
#12 0x0808ee2a in event_loop (flags=0) at event.c:468
#13 0x0808ee42 in event_dispatch () at event.c:406
#14 0x0805f889 in libeventThreadFunc (veh=0x80aacc0) at trevent.c:251
#15 0x08051497 in ThreadFunc (_t=0x80aad50) at platform.c:106
#16 0xb7d20267 in start_thread () from /lib/tls/i686/cmov/libpthread.so.0
#17 0xb7cb460e in clone () from /lib/tls/i686/cmov/libc.so.6

[charles]

turbo: I was wondering, as I look over tr_cpFileIsComplete()... did the torrent in question have any files which were zero bytes in size?

If you can track this problem down to a single torrent, could you mail me the .torrent or a link on where to get it?

Also: please don't set milestones unless you're attaching patches ;)

[charles]

If it is a zero-byte file problem, r7796 should fix it

[turbo]

No change with r7796 - still get the assertion. In different torrents each time. And no zero byte files in the either of the torrent(s).

Don't have a zero-byte file in any of my torrents...

[charles]

Hmm, too bad. At least it's repeatable for you; we can keep hammering on it. Please update to r7797 and attach the new assertion failure if/when it happens again.

[charles]

turbo: anything to report? :)

[livings124]

Do you get the same assert in the released 1.42?

[turbo]

Haven't tried 1.42 in quite a while. But r7797 did SOMETHING! I checked the differences between r7796 and r7797, and that don't explain much.

I get much higher loads, threads hang and some other things I can't quite put my finger on (my move thingie isn't doing what its supposed to).

And my primary (live) daemon is run in foreground. After a while the client (remote) can't connect (or at least, it sits there waits for data). Pressing ONE Ctrl-C in the daemon, releases the daemon and 'business as usual'. For a while at least (the time it takes the main thread to shutdown).

I'm no friend of GDB (or rather, GDB hates me! :). I've never really figured out how to follow a child thread instead of the main thread so I can't debug why...

Just saw that I need to update. I'll report back after I've done so and looked at what's changed...

[charles]

gdb isn't needed. just attach the output of the assertion failure in r7797 or higher.

[charles]

turbo: anything to report? :)

[turbo]

Unfortunatly not. It haven't crashed in days!

It hangs every now and then though. Or rather, the client can't connect. But output from the daemon indicate that it's actually working...

[charles]

Feel free to reopen this ticket if/when the problem reappears. Until then there's not much I can do about this.