Errors when installing on ARM

Bug #1030836 reported by Robie Basak on 2012-07-30
This bug affects 2 people
Affects Status Importance Assigned to Milestone
collectl (Ubuntu)

Bug Description

Setting up collectl (3.6.3-1) ...
Starting collectl: collectlpcilib: Cannot open /proc/bus/pci
lspci: Cannot find any working access method.
sh: 1: /usr/sbin/dmidecode: not found

On ARM, there is not necessarily any PCI bus or DMI facility, so there should not be an error generated here.

Note that the postinst does exit with a zero status, and AFAICT collectl does work correctly once installed. If this is not the case, please could someone affected post a test case?

ProblemType: Bug
DistroRelease: Ubuntu 12.10
Package: collectl 3.6.3-1
ProcVersionSignature: User Name 3.4.0-204.9-omap4 3.4.4
Uname: Linux 3.4.0-204-omap4 armv7l
ApportVersion: 2.4-0ubuntu6
Architecture: armhf
Date: Mon Jul 30 08:05:36 2012
PackageArchitecture: all
SourcePackage: collectl
UpgradeStatus: No upgrade log present (probably fresh install)

Robie Basak (racb) wrote :
Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in collectl (Ubuntu):
status: New → Confirmed
Trevor Robinson (scurrilous) wrote :

Yes, it seems to work correctly, despite the error messages, though I'm primarily using collectl to feed disk I/O metrics into Ganglia. I brought up the issue as primarily "fit and finish".

However, I did not look into what dmidecode is used for, so I don't know whether some part of the functionality is impacted. There may be information obtained from it that could be obtained from elsewhere on ARM/non-DMI systems.

Mark Seger (mjseger) wrote :

trevor - good catch, again! ;)

collectl uses dmidecode for only one thing - if you're doing environmental monitoring with -sE, it then uses dmidecode to get the product name which is then used in a lookup table of how to interpret the ipmitool output based on the model. so the quick answer is it's fine if it can't read it and all the stats will still be correct. It also store the product name in the header of the output files as a handy reference in case you're sharing logs with someone else or looking at them on another system, which is also harmless if it can't be written.

clearly I should be trapping the error and simply setting the product name to blanks if dmidecode is not there.

I'll get that into the next release too.


To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers