MAAS wrongly reports cannot reach PowerEdge R630 BMC during machine create and fails to commission node later on
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MAAS |
Invalid
|
Undecided
|
Unassigned |
Bug Description
Running on a cluster of 9 PowerEdge R630 with Intel Xeon E5 2640 v3
BIOS versions tested: 1.5.4; 2.5.5; 2.9.1
iDRAC version: 2.61.60.60
MAAS: 2.5.2 (7523-ge4ecbd54
Since start of the deployment, I've been seeing power-related failures during node deployment due to "BMC Busy". I decided to upgrade firmware (iDRAC and BIOS) to try remove those deployment errors.
However, when moving to other BIOS versions, nodes could not pass commissioning phase any longer.
When adding nodes with "maas create", rather than just booting, on BIOS 2.5.5 and 2.9.1, I get:
2019-04-29-20:56:00 root ERROR Command failed: machines create hostname=HOSTNAME power_type=ipmi architecture=
2019-04-29-20:56:00 root ERROR No rack controllers can access the BMC of node: HOSTNAME
I reproduced ipmi_power and ipmi_chassis_config commands as described on drivers/
ipmi-chassis-config -W opensesspriv --driver-type LAN_2_0 -h HOST_IP -u USERNAME -p PASSWORD --commit --filename ~/maas-config-test
~/maas-config-test:
Section Chassis_Boot_Flags
Boot_Device PXE
EndSection
ipmipower -W opensesspriv -D LAN_2_0 -h HOST_IP -u USERNAME -p PASSWORD --on
Both commands run OK -> freeipmi tools are working on this node
Node is correctly added to MAAS as "New" and keeps a warning on WEB UI saying no rack can reach it. Eventually, warning disappears.
The problem is that commissioning on those nodes always timeout.
When I check cloud-init configs, I can see that actually, that node received tasks for enlistment process rather than commissioning.
Steps to reproduce:
1) Set your PowerEdge R630 for the setup described above;
2) Add node using machine create command, as described
3) Run commissioning while seeing serial output of the node; machine will have "maas-enlistmen
4) Node will power off as enlistment process states but MAAS will keep waiting on commissioning until timeout
summary: |
- MAAS wrongly reports cannot reach PowerEdge R630 BMC although freeipmi - commands work fine + MAAS wrongly reports cannot reach PowerEdge R630 BMC during machine + create |
description: | updated |
summary: |
MAAS wrongly reports cannot reach PowerEdge R630 BMC during machine - create + create and fails to commission node later on |
tags: | added: cpe-onsite |
Is this still effecting you with the latest MAAS and PowerEdge R630 firmware? If so what is the return code from the IPMI command?
Can you also attach the MAAS logs?