Re: [Help-gnunet] Re: [GNUnet-developers] GNUnet 0.6.0a released

[Christian Grothoff]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Friday 10 October 2003 02:25 am, Hendrik Pagenhardt wrote:
> I suffer from repeatedly dying gnunetd ever since 0.6.0 (I use CVS, but
> the 0.6.0 release was about the time the trouble started). To identify
> the problem I tried running it in gdb, but that didn't work (some
> threads seem to hang then...). So I started a loop running "gnunetd -d"
> to give you at least the "last words" of gnunetd before dying.

Hmm. Did you use "run -d" when starting gnunetd in gdb?  Some versions of 
glibc are known to cause trouble with gdb, which in this case would clearly 
be unfortunate.  

> To be sure not to report a fixed problem I updated the CVS version
> today (Fri Oct 10 08:13:29 CEST 2003), and still I get random
> killings...

Interesting.

> before update today:
> Fri Oct 10 07:25:17 CEST 2003
> ./rungnunet.sh: line 5: 31249 Killed                  gnunetd -d
> Fri Oct 10 07:35:29 CEST 2003
> Oct 10 07:44:21 WARNING: Invalid sequence number 809 < 809, dropping rest
> of packet Oct 10 07:44:50 FATAL: knapsack error
> Oct 10 07:44:50 __BREAK__ at logging.c:220
> Fri Oct 10 07:45:00 CEST 2003
> ./rungnunet.sh: line 5:  3172 Killed                  gnunetd -d
> Fri Oct 10 07:49:03 CEST 2003
> ./rungnunet.sh: line 5:  3264 Killed                  gnunetd -d
> Fri Oct 10 07:56:44 CEST 2003
> Oct 10 07:57:52 FATAL: pthread_mutex_lock returned 22 (Invalid argument) in
> routing.c:996 Oct 10 07:57:52 __BREAK__ at logging.c:220
> Fri Oct 10 07:58:02 CEST 2003
> ./rungnunet.sh: line 5:  3473 Killed                  gnunetd -d
> Fri Oct 10 08:03:07 CEST 2003
>
> after update:
> Fri Oct 10 08:27:23 CEST 2003
> ./rungnunet.sh: line 5:  4351 Killed                  gnunetd -d
> Fri Oct 10 08:32:40 CEST 2003
> Oct 10 08:35:17 FATAL: pthread_mutex_lock returned 22 (Invalid argument) in
> routing.c:996 Oct 10 08:35:17 __BREAK__ at logging.c:220
> Fri Oct 10 08:35:27 CEST 2003
> ./rungnunet.sh: line 5:  4562 Killed                  gnunetd -d
> Fri Oct 10 08:39:02 CEST 2003
> ./rungnunet.sh: line 5:  4653 Killed                  gnunetd -d
> Fri Oct 10 08:41:17 CEST 2003
> ./rungnunet.sh: line 5:  4785 Killed                  gnunetd -d
> Fri Oct 10 08:43:39 CEST 2003
> ./rungnunet.sh: line 5:  4949 Killed                  gnunetd -d
> Fri Oct 10 09:21:13 CEST 2003
> ./rungnunet.sh: line 5:  6469 Segmentation fault      gnunetd -d
> Fri Oct 10 09:23:06 CEST 2003
>
> The knapsack error did not occur until now, but the mutex_lock and the
> seemingly reasonless "Killed" is quite common... May you can look into
> it. If you want I can file a bug report too. I'll leave gnunetd running
> supervised for today and will send the log if something else happens.

I hacked on the knapsack last night, maybe I broke something (likely).

> There's another thing... With the release of 0.6.0 we now have the
> situation, that libextractor_lower.so is no longer part of the gnunet
> package, but as far as i know, there is still no new release of
> libextractor including it (only in CVS). This will probably lead to
> newly inserted content which not as easy to locate because of the
> differing case of the keywords.

I'm just talking to Filip who is the remaining roadblock towards releasing 
0.2.6 (ok, right, I'm just waiting for his code which will port LE to OSX 
:-).

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE/htQQ9tNtMeXQLkIRAh9dAJ9I1iSfnSZmg543KiQ7mfFg905E/wCdHnSV
PAIptLkuiJFk9PQK+2nK+0A=
=R0VS
-----END PGP SIGNATURE-----