On Monday, November 24, 2014, Alan Evangelista <ala...@linux.vnet.ibm.com> wrote:
> On 11/23/2014 01:35 PM, Jeff Schroeder wrote: > >> On Saturday, November 22, 2014, Alan Evangelista < >> ala...@linux.vnet.ibm.com <mailto:ala...@linux.vnet.ibm.com>> wrote: >> >> Question 1: From the user perspective, what is the benefit of >> using Cobbler's own TFTP server, >> implemented in Python, over using inet TFTP server or another TFTP >> server ? I see it generates >> templates in RAM instead of creating files (eg boot loader >> configuration file), but I dont >> know if this translates in a performance gain. >> >> >> Great question Alan. From my experience at $previous_job running cobbler >> sync when you have thousands of records takes a very long time. With the >> built in cobbler tftp server, there is no sync after flipping the NetBoot >> boolean. I actually see that feature as a massive edge over Ohad's Foreman >> for larger installations vs waiting 60 seconds for a sync to complete. >> > > cobbler sync is naive, it writes tftp files for all systems. In Cobbler > latest code, sync between Cobbler system object > and TFTP server is done incrementally via lite_sync when netboot is > enabled/disabled in a Cobbler system > object. This decreased much the performance difference between using an > external FTP server and Cobbler's own > FTP server. This is already available in Cobbler 2.6. > Why not replace sync with lite_sync entirely? -- Text by Jeff, typos by iPhone
_______________________________________________ cobbler-devel mailing list cobbler-devel@lists.fedorahosted.org https://lists.fedorahosted.org/mailman/listinfo/cobbler-devel
