[
https://issues.apache.org/jira/browse/TS-1632?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yakov Kopel updated TS-1632:
----------------------------
Description:
In the RecDecrRawStat function (in I_RecProcess.h) there is a patch that
doesn't let the sum value to go beneath the zero value.
This can cause to a problem because the sum variable is per net-thread.
When the increase is done in one net-thread and the decrease is done in another
net-thread - the result will be 1 instead of 0.
First solution:
remove that patch
The problem with the First Solution:
If the TS-1631 fix will be in. When the TS will reset the stats, it can cause
to not positive values in the sum. this ca be when the reset is done between an
increase and decrease.
Second solution:
remove that patch
add a patch on the global sum (RecProcess.cc) and not for each net-thread
The problem with the First Solution:
It is similar to the problem with the first solution. The different is that
it won't show not negative values, but it can show lower value than the real
one.
anyone can think on a good solution?
was:
In the RecDecrRawStat function (in I_RecProcess.h) there is a patch that
doesn't let the sum value to go beneath the zero value.
This can cause to a problem because the sum variable is per net-thread.
When the increase is done in one net-thread and the decrease is done in another
net-thread - the result will be 1 instead of 0.
First solution:
remove that patch
The problem with the First Solution:
If the TS-1631 fix will be in. When the TS will reset the stats, it can cause
to not positive values in the sum. this ca be when the reset is done between an
increase and decrease.
Second solution:
remove that patch
add a patch on the global sum (RecProcess.cc) and not for each net-thread
The problem with the First Solution:
It similar to the problem with the first solution. The different is that it
won't show not negative values, but it can show lower value than the real one.
anyone can think on a good solution?
> RecDecrRawStat does not seem to work as intended
> ------------------------------------------------
>
> Key: TS-1632
> URL: https://issues.apache.org/jira/browse/TS-1632
> Project: Traffic Server
> Issue Type: Bug
> Components: Stats
> Reporter: Yakov Kopel
> Attachments: dec_stats_sol_1.diff, dec_stats_sol_2.diff
>
>
> In the RecDecrRawStat function (in I_RecProcess.h) there is a patch that
> doesn't let the sum value to go beneath the zero value.
> This can cause to a problem because the sum variable is per net-thread.
> When the increase is done in one net-thread and the decrease is done in
> another net-thread - the result will be 1 instead of 0.
> First solution:
> remove that patch
> The problem with the First Solution:
> If the TS-1631 fix will be in. When the TS will reset the stats, it can
> cause to not positive values in the sum. this ca be when the reset is done
> between an increase and decrease.
> Second solution:
> remove that patch
> add a patch on the global sum (RecProcess.cc) and not for each net-thread
> The problem with the First Solution:
> It is similar to the problem with the first solution. The different is
> that it won't show not negative values, but it can show lower value than the
> real one.
> anyone can think on a good solution?
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
