Current beaker-system-scan determines vendor and model from some unknown place? lshw does not appear to do this. Need to figure out where these strings come from in smolt and whether lshw can produce the same. For example on a KVM guest: Vendor: Red Hat Model: KVM RHEL 6.0.0 PC
(In reply to Dan Callaghan from comment #0) > Current beaker-system-scan determines vendor and model from some unknown > place? lshw does not appear to do this. Need to figure out where these > strings come from in smolt and whether lshw can produce the same. smolt gets it from '/org/freedesktop/Hal/devices/computer'. Let's see where else that is available from.
(In reply to Amit Saha from comment #1) Might be worth figuring out where HAL itself was getting that info from. In many cases udev will now have the same info.
(In reply to Dan Callaghan from comment #2) > (In reply to Amit Saha from comment #1) > > Might be worth figuring out where HAL itself was getting that info from. In > many cases udev will now have the same info. HAL gets it from SMBIOS which is also used by lshw. The XML parsing was not correct in beaker-system-scan: http://gerrit.beaker-project.org/#/c/4188/ I will need to verify it on the other systems.
I think we will need to patch lshw to extract the vendor correctly on POWER. This is clearly not right: 'model': 'IBM,8286-42A', 'vendor': None Also 'JS20' would be a more useful model string than '8842-P3A' (similarly 'P750E' would be more useful than '8233-E8B') but I guess we can live with it.
(In reply to Dan Callaghan from comment #12) > I think we will need to patch lshw to extract the vendor correctly on POWER. > This is clearly not right: > > 'model': 'IBM,8286-42A', > 'vendor': None lshw gets the model value from: # cat /proc/device-tree/model IBM,8286-42A 8286-42A seems to be the model here: http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=an&subtype=ca&appname=gpateam&supplier=897&letternum=ENUS114-059 What do you think of just splitting on the , and set vendor to IBM and model to 8286-42A ?
(In reply to Amit Saha from comment #16) > What do you think of just splitting on the , and set vendor to IBM and model > to 8286-42A ? The IEEE 1275 spec doesn't prescribe any particular meaning for the "model" property beyond: Standard property name to define a manufacturer's model number. However it seems to be a convention in the spec to use "vendor,model" separated by a comma and I guess that is followed in the model property as well. So I think we could update lshw to split on , if present.
(In reply to Dan Callaghan from comment #12) > I think we will need to patch lshw to extract the vendor correctly on POWER. > This is clearly not right: > > 'model': 'IBM,8286-42A', > 'vendor': None > > Also 'JS20' would be a more useful model string than '8842-P3A' (similarly > 'P750E' would be more useful than '8233-E8B') but I guess we can live with > it. http://gerrit.beaker-project.org/#/c/4193/1
Verify passed on beaker-system-scan-1.6-1.git.19.1c8487e.el6
Beaker 21.0 has been released.