-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.18.0
-
Important
-
None
-
Rejected
-
False
-
When testing the agent-based installer using iSCSI with dev-scripts https://github.com/openshift-metal3/dev-scripts/pull/1727 it was found that the installer was not able to complete the installation when using multiple hosts. This same problem did not appear when using SNO.
The iscsi session from all the hosts work do their targets fine until coreos-installer is run, at which time (before reboot) the connection to the target is lost and the coreos-installer fails
Jan 09 16:12:23 master-1 kernel: session1: session recovery timed out after 120 secs Jan 09 16:12:23 master-1 kernel: sd 7:0:0:0: rejecting I/O to offline device Jan 09 16:12:23 master-1 kernel: I/O error, dev sdb, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2 Jan 09 16:12:23 master-1 installer[2937]: time="2025-01-09T16:12:23Z" level=info msg="\nError: syncing data to disk\n\nCaused by:\n Input/output error (os error 5)\n\nResetting partition table\n" Jan 09 16:12:23 master-1 installer[2937]: time="2025-01-09T16:12:23Z" level=warning msg="Retrying after error: failed executing /usr/bin/nsenter [--target 1 --cgroup --mount --ipc --pid -- coreos-installer install --insecure -i /opt/install-dir/master-f3c24588-2129-483f-9dfb-8a8fe332a4bf.ign --append-karg rd.iscsi.firmware=1 --append-karg ip=enp6s0:dhcp --copy-network /dev/sdb], Error exit status 1, LastOutput \"Error: syncing data to disk\n\nCaused by:\n Input/output error (os error 5)\n\nResetting partition table\nError: syncing partition table to disk\n\nCaused by:\n Input/output error (os error 5)\""
On the host it can be seen that session shows as logged out
Iface Name: default Iface Transport: tcp Iface Initiatorname: iqn.2023-01.com.example:master-1 Iface IPaddress: [default] Iface HWaddress: default Iface Netdev: default SID: 1 iSCSI Connection State: Unknown iSCSI Session State: FREE Internal iscsid Session State: Unknown
The problem occurs because the iscsid service is not running. If it is started by iscsadm then coreos-installer can successfully write the image to disk.