[Condor-users] CondorView Server Bug?

[<Greg.Hitchen@xxxxxxxx>]

Hi All

 

We've recently been trying to sort out a "glitch" in our viewserver graphs that has appeared this month.

It's a bit hard to explain but basically we think we've narrowed it down to the make_machine_stats

script. We think what is happening is that if the time period that the make_machine_stats script is

spanning contains a CHANGE in your resource group list (using condor_stats -resgrouplist) then the

machine numbers for each category will get put into the incorrect groupings after that change happens.

 

e.g. early in August we had

 

INTEL/LINUX

INTEL/WINNT50

INTEL/WINNT51

Total

INTEL/WINNT52

INTEL/WINNT61

X86_64/LINUX

 

Later in August we had

 

INTEL/WINNT50
INTEL/WINNT51
Total
INTEL/WINNT52
INTEL/WINNT61
X86_64/LINUX

i.e. no INTEL/LINUX group (this had been one test ubuntu VM).

 

What happens is that up until the change the "Total" graph is OK but after that is is actually the WINN52

graph, the WINNT51 becomes the "Total", etc., etc.

 

Has anyone else seen this? or can confirm this?

 

Thanks.

 

Cheers

 

Greg