[Pvfs2-developers] question about defines common to server and client

Phil Carns carns at mcs.anl.gov
Thu Feb 21 10:36:21 EST 2008


Sure, I think it would be fine to move it somewhere shared.


There is kind of a larger scale problem here too, but it would be 
tedious to fix.  If you take the client job bmi timeout, for example, 
there are actually 4 places where defaults are defined:

- client-state-machine.h
- pvfs2-server.h
- pvfs2-genconfig
- server-config.c

For all practical purposes the genconfig is what wins since we emit 
almost all of these values by default.  Its rather confusing to know 
what to do with the others, though, and which ones get used where :)

The client job timeouts are a slightly special case because we need to 
know a timeout to use for the initial getconfig/mount operations before 
the config file has been retrieved.  It would also be nice if 
getconfig/mount doesn't take an eternity to time out, because most 
admins would like for a mount or pvfs2-ping to fail reasonably quickly 
even if the global timeout values have been boosted for other reasons.

I've always kind of wished that the timeout for getconfig (for all 
tools) was explicitly defined as a separate value rather than using the 
general purpose job timeout defaults.

-Phil

Walter B. Ligon III wrote:
> These defines are in both client-state-machine.h and pvfs2-server.h
> 
> Can they be moved to header that is already included in both places?
> 
> Walt
> 
> /* Default client timeout in seconds used to set the timeout for jobs that
>  * send or receive request messages.
>  */
> #define PVFS2_CLIENT_JOB_BMI_TIMEOUT_DEFAULT 30
> 
> 
> /* Default number of times to retry restartable client operations. */
> #define PVFS2_CLIENT_RETRY_LIMIT_DEFAULT  (5)
> 
> 
> /* Default number of milliseconds to delay before retries */
> #define PVFS2_CLIENT_RETRY_DELAY_MS_DEFAULT  2000
> 



More information about the Pvfs2-developers mailing list