On 30/07/09 11:55, Wojtek Meler wrote: > Fabio M. Di Nitto pisze: >> testcpg is not endian clean. That's why you see them backwards. >> > I thought that corosync handles endianess in corosync daemon, and > clients - like testcpg - > don't need to worry about it. Is it true? > I think that problem is in nodeid interpretation. It is initialized with > node IP (always network order?), > but then handled like all ints (swab32).So when it is transmitted in > cluster you can't guess > if you should swab it again even on architectures that don't need it > (sparc)... > quite common mistake.
If you don't set the nodeid then it defaults to the IP address. But it's then an int, not an IP address, as far as corosync is concerned. The fact that it was an IP address is irrelevant, it's just a way of getting a unique node number. The error here is in testcpg passing it into inet_ntoa. In many clusters - especially those using the Red Hat stack, the nodeIDs will be small integers, and the output from that will be rubbish. Chrissie _______________________________________________ Openais mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/openais
