[Pgcluster-general] Cybercluster possible Bug

[Lia Domide]

Hello everybody,

 

I am testing now cybercluster.

A problem in "cluster_table.c" (PGRscan_cluster) appears when more than 2 DB
nodes are registered to a load-balancer. The code above uses "cnt" variable
from 0 to "<= ClusterNum", and this variable is incremented twice per each
iteration. For <=2 nodes is fine, but for >=3 it omits the last nodes.

ClusterTbl * PGRscan_cluster(void){

....

    cnt =0

    while ((cnt <= ClusterNum) && (ptr->useFlag != TBL_END)){    

         ...

          cnt ++;

          if..

          cnt ++;

           ...

   } // end while

}

One of the "annoying" operations: "cnt++" should be dropped, I think.

 

 Have a nice day,

   Lia Domide.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pgfoundry.org/pipermail/pgcluster-general/attachments/20080125/af9be14a/attachment.html