On 08/19/05 10:03, you wrote :
Secondly, there is now an 'unmagic' number, 1290, which is set in HostTraffic during the start of the delete process. Because of the timing of the messages (between selection and completion of the purge), I'm suspicious that we are reusing deleted entries and/or have a pointer to them during the delete process. So if your messages become:
Aug 17 06:46:34 gbp ntop[23566]: [MSGID8850278] **WARNING** Error: bad magic number [expected=1968/real=1290][deviceId=0] getFirstHost()[address.c/54]
That will tell us the problem is reusing deleted entries. If it's still zero, then the problem is something else.
Guess what: after a few (probably unrelated) crashes, the “bad magic number” message came back:
Aug 24 22:53:07 gbp ntop[1808]: [MSGID8931710] RRD: Cycle 153 ended, 59 RRDs updated
Aug 24 22:53:20 gbp ntop[1808]: [MSGID0825709] IDLE_PURGE: Device 0 [eth1] FINISHED selection, 1747 [out of 10213] hosts selected
Aug 24 22:53:21 gbp ntop[1808]: [MSGID8477291] IDLE_PURGE: Device 0 [eth1]: 1747/10212 hosts deleted, elapsed time is 0.051907 seconds (0.000030 per host)
Aug 24 22:53:28 gbp ntop[1808]: [MSGID8757584] SFP: Ending fingerprint scan cycle 304 - checked 131, resolved 0
Aug 24 22:55:21 gbp ntop[1808]: [MSGID0825709] IDLE_PURGE: Device 0 [eth1] FINISHED selection, 1409 [out of 9895] hosts selected
Aug 24 22:55:21 gbp ntop[1808]: [MSGID8477291] IDLE_PURGE: Device 0 [eth1]: 1409/9894 hosts deleted, elapsed time is 0.051273 seconds (0.000036 per host)
Aug 24 22:55:59 gbp ntop[1808]: [MSGID8757584] SFP: Ending fingerprint scan cycle 305 - checked 126, resolved 0
Aug 24 22:56:53 gbp ntop[1808]: [MSGID8781395] **ERROR** Bad magic number (expected=1968/real=10248) [deviceId=0] lookupHost()[pbuf.c/1088]
I grep’ed for the unmagic number (1920) and found it in globals-defines.h but didn’t find any reference to “10248”. Where does that number come from?
This time all ntop processes have died so there is no process left to run gdb on.
Regards,
Marc.
--
Marc Mazuhelli
Computer security analyst
Service des technologies de l'information
Université de Sherbrooke
_______________________________________________ Ntop-dev mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop-dev
