Current inventory data records "GenuineIntel" which is also the CPUID vendor string I believe. lshw is probably mapping this to some kind of human friendly name. beaker-system-scan needs to preserve the original values here as they are important for existing host filters. Presumably same issue exists with "AuthenticAMD"?
(In reply to Dan Callaghan from comment #0) > Current inventory data records "GenuineIntel" which is also the CPUID vendor > string I believe. lshw is probably mapping this to some kind of human > friendly name. > > beaker-system-scan needs to preserve the original values here as they are > important for existing host filters. lshw does indeed check for those strings and replace them by human friendly versions. Do we just do the reverse mapping in beaker-system-scan? > > Presumably same issue exists with "AuthenticAMD"? Indeed.
(In reply to Amit Saha from comment #2) > lshw does indeed check for those strings and replace them by human friendly > versions. Do we just do the reverse mapping in beaker-system-scan? It would be nice if we can also expose the raw CPUID string "GenuineIntel" in the lshw XML somewhere in a backwards compatible way. It seems generally useful to have that. Then beaker-system-scan can look for that in the XML and just ignore the human-friendly name. Maybe like this: <node id="cpu" ... <vendor>Intel Corp.</vendor> <vendorid>GenuineIntel</vendorid> Failing that, just mapping the human names back to the CPUID strings in beaker-system-scan is an okay choice.
(In reply to Dan Callaghan from comment #3) > (In reply to Amit Saha from comment #2) > > lshw does indeed check for those strings and replace them by human friendly > > versions. Do we just do the reverse mapping in beaker-system-scan? > > It would be nice if we can also expose the raw CPUID string "GenuineIntel" > in the lshw XML somewhere in a backwards compatible way. It seems generally > useful to have that. Then beaker-system-scan can look for that in the XML > and just ignore the human-friendly name. > > Maybe like this: > > <node id="cpu" ... > <vendor>Intel Corp.</vendor> > <vendorid>GenuineIntel</vendorid> Good idea: http://gerrit.beaker-project.org/#/c/4187/1 Filed an upstream ticket too with the patch, but it's moderated, so not visible yet. > > Failing that, just mapping the human names back to the CPUID strings in > beaker-system-scan is an okay choice.
re: comment 1: I've been doing some testing with lshw on RHEL6 and 7 with the lshw v-r that is shipped on these two releases. The Model Name string as presented in /proc/cpuinfo is listed under a "version" field in lshw. I like just exposing the raw vendor_id field though, I'll clone this for RHEL 6.7/7.2
(In reply to Mike Gahagan from comment #5) > re: comment 1: > > I like just exposing the raw vendor_id field though, I'll clone this for > RHEL 6.7/7.2 Upstream isn't interested in doing this: http://www.ezix.org/project/ticket/683 So, we have just decided to do a renaming in beaker-system-scan which is the tool we use for gathering inventory on beaker systems: https://git.beaker-project.org/cgit/beaker-system-scan/
(In reply to Amit Saha from comment #6) > (In reply to Mike Gahagan from comment #5) > > re: comment 1: > > > > I like just exposing the raw vendor_id field though, I'll clone this for > > RHEL 6.7/7.2 > > Upstream isn't interested in doing this: > http://www.ezix.org/project/ticket/683 > > So, we have just decided to do a renaming in beaker-system-scan which is the > tool we use for gathering inventory on beaker systems: > https://git.beaker-project.org/cgit/beaker-system-scan/ Patch: http://gerrit.beaker-project.org/#/c/4191/
(In reply to Mike Gahagan from comment #5) > I've been doing some testing with lshw on RHEL6 and 7 with the lshw v-r that > is shipped on these two releases. Huh... I did not realise that lshw was actually shipped in any RHEL releases. We have been including it in the Beaker harness repos (with various patches applied as needed) for a while now. The good news is that we have of course been upstreaming all our bug reports and patches, so as long as RHEL is willing to rebase lshw the fixes should all eventually make their way into the shipped version of lshw.
Verification instructions: 1. Reserve a Intel x86_64 box with RHEL 7 2. Install beaker-system-scan using "yum install beaker-system-scan" 3. Run, "beaker-system-scan -d" See that CPU vendor is listed as "GenuineIntel"
Beaker 21.0 has been released.