Monday morning saw half the DPO failing to boot into linux, which caused some minor disruption (fortunately the practicals could use x sessions on x-access from windows to work around).
The problem appeared to be that the PXE boot session wasn’t getting any DHCP replies, even though the logs on the DHCP servers showed the queries coming in and the replies being sent. This cause them to time out, boot from disk (Windows) and carry on. Windows would also DHCP for an IP address, but probably was carrying on happily using the one it had previously had – though whether that lease would eventually expire remains an unknown.
The problem was cured by resetting the switch which serves that half of the DPO. Because the DHCP servers are not on the same VLAN as the clients, the switch/router fabric has to have a DHCP helper configured to forward the broadcast DHCP requests to the servers and to handle and rebroadcast the replies. The first part was working, the second was not. After the switch was rebooted, all started to work again as normal with the DPO clients getting a DHCP reply and dropping into the PXE prompt to allow the choice of linux or windows boot.
Leave a Reply
You must be logged in to post a comment.