* Aaron Stone ([EMAIL PROTECTED]) [340218 21:34]: > serverchild.c,active_child_sig_handler: got signal [Alarm clock] > *** glibc detected *** corrupted double-linked list: 0x0805f028 *** > pool.c,child_register: register child [19712] > pool.c,manage_spare_children: children [1/1], spares [1 (0 - 0)] > > These all came from dbmail-timsieved from the console, and connecting to > it with gsieve (a little PyGTK managesieve client). I left gsieve open > and went to sleep, and woke up to find that the connection had timed out > (of course), with those messages on the console.
Last time we had these problems was a valuable lesson. Fat chance it's the trace calls in the pool code. All code paths starting in serverchild.c,active_child_sig_handler need to be reviewed for signal safety. I don't have the list of safe calls handy, but the trace calls in the pool code are very much suspect. Goood thing you caught this. Could you file a bug please? -- ________________________________________________________________ Paul Stevens mailto:[EMAIL PROTECTED] NET FACILITIES GROUP PGP: finger [EMAIL PROTECTED] The Netherlands________________________________http://www.nfg.nl
