Type |
New Feature
|
Priority |
Normal
|
Assignee | |
RedmineID |
4504
|
Version |
tigase-server-7.1.0
|
Spent time |
13h 45m
|
Related
Issue Votes (0)
Watchers (0)
Type |
New Feature
|
Priority |
Normal
|
Assignee | |
RedmineID |
4504
|
Version |
tigase-server-7.1.0
|
Spent time |
13h 45m
|
While looking at the statistics (and sizes of them) I've noticed lot's of similar entries: @cl-comp/Processed packets thread…@. This is especially noticeable with higher number of CPU/cores. With only 4 of them we get roughly 2x16 entries for majority of components, but 2x32 entries for cl-comp, and 2x64 for sess-man (see below). With 16 cores this grows even more - uses more memory, size on disk and actually makes it more difficult to look at the statistics.
I think that turning it into array (like @cl-comp/Last day disconnects@) would be a good idea:
In addition, the biggest advantage of this information is to know whether the load is spread evenly across the threads, so we could add additional statistic with information about the distribution and/or if there are any outliers.
Thoughts?