You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Dean Gaudet <dg...@arctic.org> on 1997/11/04 22:01:27 UTC
"tomahawk" update
I'm tenatively using the name tomahawk, thanks :)
- "long lost child came home" after a USR1 with a piped log child, due to
improper callback logic I think
- TransferLog "| bin/aplogger logs/logger.pid >> logs/access_log" running
for 24 hours produced some interesting results. The aplogger never
died, but the /bin/sh which was running it did. Since /bin/sh is
the pid that apache is watching, it would spawn another logging
child. After 24 hours I had about 40 aploggers running... things
still work OK (because of how pipes and such work), it just sucks.
I'm not sure yet why /bin/sh was dying. I've replaced it with:
TransferLog "| exec bin/aplogger logs/logger.pid >> logs/access_log"
Oh hmm. I wonder if it's because Apache thought it was write blocked.
Hmm. I'll add logging code to look for this.
Note to self: aplogger needs to periodically test getppid() to make
sure it hasn't lost its parent; just a safe-guard.
- listenwrap seems to be producing "bind: address already in use"; unsure
yet if it's a networking bug in my kernel, or if I'm tripping one of
the networking security checks added after the NFS port 2049 fiasco
security bug a while back. In any event this goes on until all old
sockets are destroyed or reach FIN_WAIT_2.
Dean