[lkml]   [1999]   [Feb]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: fsync on large files
Olaf Titz <> writes:
> > The real fix for this problem would be to use a syslogd which uses datagram
> > sockets instead. (Both libc5 and glibc understand them.)
> Which means, when syslogd is really overloaded, silently throwing away
> half of the messages. Hmph.

The current way of doing things is worse, because I can cause syslogd to
silently throw away _all_ messages (from processes which haven't syslogged
anything before).

Simply open /dev/log until you run out of file descriptors. Repeat that
until you run out of process slots. If that still falls short (some sites
do have restrictive ulimits), connect to an inetd-based service and cause
its server to syslog something (but not die); I assume there are servers
which can be convinced to do this.

Repeat until successful.

If you're concerned about lossage, then make unix-domain sockets nonblocking.
Shouldn't be too difficult...

Matthias Urlichs | noris network GmbH | | ICQ: 20193661
The quote was selected randomly. Really. |
In the factory, we make cosmetics. In the store, we sell hope.
-- Charles Revson
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:50    [from the cache]
©2003-2014 Jasper Spaans. hosted at Digital Ocean