After your VOS or OpenVOS module has been up for a while (a few months or more), you may notice that some of the metering values displayed by various analyze_system requests have gone negative, or some of the percentages are way off. This is caused by the underlying integer counters overflowing; when this happens, they “wrap around” to the largest negative number and start counting up through the negative numbers.  Eventually, if you wait long enough, they will become positive again.  But since the counters are incremented at various times, it is guaranteed that only some counters will be negative, while others remain positive. Hence, the meaningless output of negative values.

The solution is simple: reset the meters before attempting to reference them.  The “-reset” option to the various analyze_system metering requests saves a copy of the meters in a file named as_meter_file in your home directory.  The next time that you execute the metering request, these saved values are subtracted from the current values, so that the data represents just the activities that took place since the meters were reset.

If you want to “unreset” the meters, simply rename or delete as_meter_file.

Here is an example of using the analyze_system cache_meters on a module that has been up for a few months. The first use of cache_meters displays data that is inconsistent. The second use displays data that is accurate.

                              Hits              Misses              Total
File       Data      1112734947/ 93.81%   73384889/  6.19% 1288214193/279.61%
           Indirect   648741913/ 98.00%   13221914/  2.00%  662389071/143.77%
Index      Data        56569823/ 95.60%    2605721/  4.40%   59182760/ 12.85%
           Indirect       23459/ 97.40%        625/  2.60%      24203/  0.00%
Directory  Data      **********/******%   12033330/******% **********/******%
           Indirect    17025043/ 99.75%      42070/  0.25%   17068785/  3.70%
Totals                253698831/ 71.47%  101288549/ 28.53%  460716273
as:  cache_meters -reset
as:  cache_meters
cache_meter normal   %es#m105                   0:00:37    10-07-19 16:04:18
Metering time:      0:00:37
___FCKsi___5                              Hits              Misses              Total
File       Data           10124/ 52.27%       9246/ 47.73%      22052/ 47.33%
           Indirect       17355/ 98.92%        190/  1.08%      17547/ 37.66%
Index      Data              10/ 90.91%          1/  9.09%         11/  0.02%
           Indirect           0/  0.00%          0/  0.00%          0/  0.00%
Directory  Data            6748/ 96.69%        231/  3.31%       6979/ 14.98%
           Indirect           6/100.00%          0/  0.00%          6/  0.01%
Totals                    34243/ 77.98%       9668/ 22.02%      46595
as:
___FCKsi___5___FCKpd___5
___FCKpd___5

I hope this helps clarify the situation.