[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Linux-cluster] [PATCH 0/4] fence



On Fri, Feb 18, 2005 at 12:01:35PM +0800, David Teigland wrote:
> On Thu, Feb 17, 2005 at 07:35:34PM +0100, Bastian Blank wrote:
> > Patches:
> > 1. Use bools instead of crude defines.
> Interesting, but I think I still prefer using int's and defines.  I like
> the option of doing other things (-1/0/1) even if we're only doing T/F
> now.  Also, there's no /usr/include/stdbool.h on my machine.

Try to find it in the include dir of your compiler, stdbool.h is not
implementable by the libc, which puts its includes into /susr/include.

If you want to use ints, please wrap them into an emun.

> > 2. Remove GNU-isms.
> You'll have to pardon my ignorance about GNU-isms; could you explain the
> shortcoming of the current method, or point me to a discussion?  The
> existing code looks nicer to me, but I'm happy to learn something new.
> 
> For reference, existing:
> #define die(fmt, args...) \
                   ^^^^^^^
> do \
> { \
> 	fprintf(stderr, "%s: ", prog_name); \
> 	fprintf(stderr, fmt "\n", ##args); \
> 	syslog(LOG_ERR, fmt, ##args); \
                             ^^^^^^
> 	exit(EXIT_FAILURE); \
> } \
> while (0)

The variable argument list is a GNU-ism.

> > 4. Wait for the join complete event in fenced.
> Or, I just thought of another method.  fence_tool's -w handling could
> could read fenced's unix socket and wait until it sees "finish:".  See
> fence_tool.c:do_monitor().  do_monitor("finish:") would return when it
> sees a line matching "finish:".
> We could also use this method to allow "fence_tool leave -w".

Hmm, lets think about it.

Bastian

-- 
If there are self-made purgatories, then we all have to live in them.
		-- Spock, "This Side of Paradise", stardate 3417.7

Attachment: signature.asc
Description: Digital signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]