[Help-gnunet] gnunetd suddenly very unstable

[Christian Drechsler]

hey there!

all the time i had set my indirection table size to 131072, and gnunetd
0.4.6 was very, very stable - never crashed at all, to be precise. now i
decided to have some more free memory, ;-) so i wanted to decrease the
size again. first i set it to 8092 - as proposed in config.h - which isn't
a power of two. ;-) gnunetd crashed shortly after startup just when saying
that the indirection table was full - no other message. then i remembered
8192 is the real power of 2, so i changed it to that value. gnunetd
survived for some seconds  after stating that the indirection table was
full but then crashed again - all within one minute from startup. so i set
the value to 16384. as i stopped gnunet-download the "query load" was not
_that_ high; gnunetd never said it reached the end of its indirection
table. nevertheless it died after about 30-50 seconds without telling me
why. somebody seems to download something of which i have many parts here,
so there are many local lookups succeeding at the moment. that seems to be
too high a "load" for gnunetd. last message in the logs (level 9) is
always:

Sep 25 00:39:37 CONNECTION: done appending message.

increasing the indirection table size to 32768 didn't help, either. i'll
try 65536 now ...

well, the same again:

Sep 25 00:44:59 CONNECTION: done appending message.

last sign of a lively gnunetd ... now what will happen if i set it to the
original value?

shit, same again. :-(

what's wrong? it worked all the time, didn't it? well, i'll set the log
level back to 6 and the indirection table size to 16384 - and try to keep
gnunetd going - with a cron script, if needed.

anybody experiencing similar stuff?

regards, zottel

ps: now (with log level still 9) the last message was:

Sep 25 00:56:16 ROUTING: Query is pending, not forwarded, no local lookup

what on earth is wrong, suddenly?