Jonathan Houser wrote:
It's been a while since I've looked at this code, but if I build Kannel with assertions turned off, will it stop shutting down with panics or will it instead crash violently? I ask because I have always had issues with wapbox hitting asserts and shutting down, thus disabling WAP 1 browsing and PPG and whatever else that entails. That'd be fine in a test environment as then I'd know something went wrong, but this is production with live customers. The asserts generally look like things that shouldn't be fatal for wapbox -- the latest a few minutes ago was:
2005-02-15 12:04:30 [27939] [8] PANIC: gwlib/octstr.c:2340: seems_valid_real: Assertion `ostr->len + 1 <= ostr->size' failed. (Called from gwlib/octstr.c:317:octstr_dest
roy.)
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox(gw_panic+0x19a) [0x80d03f7]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox [0x80d816d]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox(octstr_destroy+0x2f) [0x80d28d1]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox(octstr_destroy_item+0x11) [0x80d2905]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox(list_destroy+0x35) [0x80ce555]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox(http_destroy_headers+0x1e) [0x80cc567]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox [0x80c80f2]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox(http_receive_result+0xb4) [0x80ca691]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox [0x805c21d]
2005-02-15 12:04:30 [27939] [8] PANIC: /usr/local/sbin/wapbox [0x80c648a]
2005-02-15 12:04:30 [27939] [8] PANIC: /lib/libpthread.so.0 [0x400a4dd6]
2005-02-15 12:04:30 [27939] [8] PANIC: /lib/libc.so.6(__clone+0x5a) [0x4037a9aa]
I mean sure it's horrible that the lengths were off, but it's not so horrible that we should kill all WAP/PPG traffic until a monitoring script recovers the system. I'm guessing ./configure --disable-assertions would fix this, but I want to check as this *is* production. TIA. :)
ok, this assertions means, ocstr_destroy() seems to destroy something that is not an Octstr nor a NULL pointer. Obviously we have a defined Octstr pointer here that is not initialized, but tried to be destroyed.
Can you provide us some more lines before the first PANIC?!
And also please use 'addr2line' to resolve the source code lines of the backtrace addresses, see 'man addr2line' for details on how to use.
Stipe
mailto:stolj_{at}_wapme.de ------------------------------------------------------------------- Wapme Systems AG
Vogelsanger Weg 80 40470 D�sseldorf, NRW, Germany
phone: +49.211.74845.0 fax: +49.211.74845.299
mailto:info_{at}_wapme-systems.de http://www.wapme-systems.de/ -------------------------------------------------------------------
