Strange sar output

I was reviewing yesterday's sar file and came across this strange output! What in the world? Any reason why there's output like that?

SunOS unixbox 5.10 Generic_144488-07 sun4v sparc SUNW,T5240 Solaris

00:00:58   device        %busy   avque   r+w/s  blks/s  avwait  avserv


11:20:01   vdc0             35     2.6      27    5292     0.0    96.1
           vdc1              0     0.0       0       0     0.0     0.0
           vdc2              0     0.0       0       0     0.0     0.0
           vdc3             66     3.0     230   52338     0.0    12.8
           vdc4             64     2.9     230   52361     0.0    12.4

11:40:02   vdc0             25     1.5      24    4464     0.0    63.1
           vdc1              0     0.0       0       0     0.0     0.0
           vdc2              0     0.0       0       0     0.0     0.1
           vdc3             58     2.4     214   48830     0.0    11.4
           vdc4             56     2.3     213   48821     0.0    11.0

12:00:52   vdc0             31     2.1      27    4906     0.0    77.9
           vdc1              0     0.0       0       0     0.0     0.0
           vdc2              0     0.0       0       0     0.0     0.0
           vdc3             61     2.5     215   48894     0.0    11.8
           vdc4             60     2.5     217   49059     0.0    11.4

12:40:02   vdc0         -24893 -1460.9 1773679 15026318146864    -0.0    -0.8
           vdc1           -533    -9.3 1790730 15026272364076    -0.0    -0.0
           vdc2           -421    -7.2 1790727 15026270415816    -0.0    -0.0
           vdc3              0     0.0       0       0     0.0    19.0
           vdc4              0     0.0       0       0     0.0    26.6

13:00:00   vdc0              6     0.4       5     697     0.0    84.5
           vdc1              0     0.0       0       0     0.0     0.0
           vdc2              0     0.0       0       0     0.0     0.0
           vdc3              0     0.0       0       0     0.0     0.0
           vdc4              0     0.0       0       0     0.0     0.0

It would help to know what those statistics are. But in general there are counters in the kernel that record interesting events. If the counter is at, say, 8000, at one time and then 8200 ten seconds later, we know that 200 things happened in the last 10 seconds. But what is the counter is a 16 bit counter and it is at 65000 at the first reading and then 125 ten seconds later? The counter rolled. This screws up the math.

Thanks for your response. I edited the first post to include a header. What prompted my investigation of the sar files was, as you can see, long disk service times on vdc0.

I understand what you wrote about counter resets, but shouldn't the program take into account the integer boundaries? Anyway, I grepped through the rest of the sar files. This particular occurrence was the only example.

Of course it should.

Looks like bug 4716896 which doesn't seem to be fixed.

1 Like