I discovered reverse DNS seem not to work properly, because testing « host <compute_node_ip> » says : not found. The corresponding named db file only contains the master node, is it supposed to also have all compute nodes ? Isn’t the DNS server configuration handled by xcat ? The site dns attributes seem to be correct.
Best regards Antoine Huette HPC engineer Bechtle > Le 24 juin 2020 à 10:55, Huette, Antoine <antoine.hue...@bechtle.com> a écrit > : > > Hello, > > We have some issues deploying compute nodes on a CentOS 8.1 cluster with xCAT > 2.16. > > First, OS installation takes a lot of time, in the xcat.log file there are a > lot of « updateflag.awk : retrying flag update » messages after the > confignetwork postscript execution. They last for about 17 minutes. > > Then, there are also issues when attempting to deploy an osimage to the > compute nodes. Some attempts fail to happen, the node will just boot from > hard drive without reason. When installing multiple nodes at once, a lot of > them will fail to boot the installer environment, and will instead boot to > emergency mode. This issue is resolved by rebooting the faulty nodes until > the installation is successful. > > Does anyone know what could be the cause for these issues ? It looks like > there is a network problem here but I’m not sure where to look at... > > By the way, the compute nodes are configured to have bonded interfaces on the > main internal network that is also used for OS deployment. Are there any > considerations or special configuration to be done in that case ? The bonding > is successfully configured and fonctional at boot. > > > Best regards, > > Antoine Huette > HPC engineer > Bechtle _______________________________________________ xCAT-user mailing list xCAT-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xcat-user
