[RC5] AIX RC5 client
j-zbiciak1 at ti.com
Tue Jul 21 09:31:51 EDT 1998
'simon at geordie.demon.co.uk' said previously:
[starting 8 clients sharing buff-in causes them to crack same block]
| You need to make sure that they aren't using a checkpoint file, or make
| sure each checkpoint is pointed to a diferent file. Other than that, is
| should be OK.
This isn't necessarily true.
When I was sharing buff-in files over NFS on some Solaris machines, if
two clients grabbed a block at the same time, sometimes they would grab
the same block. There seemed to be some sort of race condition between
acquiring a lock on two different hosts for the same file. Either
that, or the locking mechanism they are using doesn't work for NFS.
With this setup, I was duping a very large proportion of my blocks done
on these hosts (15-20%, IIRC). Not good. (Incidentally, since I was
using a fair number of hosts, I had checkpoint files turned off to keep
NFS traffic to a minimum, so that wasn't the problem.)
The only cure for me was to use separate buff-in files (a shared
buff-out is OK as long as only one client flushes).
If I were doing this today, I would set up a PProxy (now that they can
talk through firewalls) and set all the clients to buffer to memory,
buffering only 2 blocks (one being cracked, and one waiting "on
+------- Joseph Zbiciak ------+
|- - - j-zbiciak1 at ti.com - - -| "The truth of a proposition has nothing
| -Texas Instruments, Dallas- | to do with its credibility.
|- - #include <disclaim.h> - -| And vice versa." -- Lazarus Long
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest
More information about the rc5