-
Notifications
You must be signed in to change notification settings - Fork 467
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Stats prepare for multiple queues per destdrv #2302
Stats prepare for multiple queues per destdrv #2302
Conversation
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
@bazsi: I think this is what we've talked about
( unit tests are missing, will added after you tried it) |
Build FAILURE |
…stered Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
754837b
to
b3768f9
Compare
Build FAILURE |
It has to be added to the memory_usage; -> and make it possible to remove * memory_usage_qout_initial_value and * memory_usage_overflow_initial_value from LoqQueue Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
These variables should never be part of LogQueue. Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
Build FAILURE |
1 similar comment
Build FAILURE |
50224cb
to
1472845
Compare
Build FAILURE |
1472845
to
db5c21a
Compare
Build FAILURE |
db5c21a
to
13a84a7
Compare
@bazsi: I checked again why tests has failed, and I think in case of diskq we cannot rely on a cached queued_messages value after the reload. I hope that it solve all the issues. |
Build FAILURE |
…ring ctrs The cached value for queued_messages cannot be use after a reload. When a LogQueue instance is found the very first thing what are doing with that is unref (in case of diskq). On the other hand, we need to track queued_messages to ensure that the StatsCounter is 0 after a deinit (without the autoreset feature). Signed-off-by: Laszlo Budai <stentor.bgyk@gmail.com>
13a84a7
to
ff79236
Compare
Build SUCCESS |
I have now rebased my patches against this series, and code-wise it seems to have worked ok. I am yet to do functional testing. At the end I've pushed the DROPPED counter down into log_queue_register_stats_counters(), the This way both the driver and the associated LogQueue would grab the same counter, the only This is the patch I am referring to: be8a4bb |
@bazsi: Should I add this to my PR? (if it causes conflict, I'd not touch it...) |
I have pushed that change to the bottom of my branch, so please do cherry-pick it over. I still have some rebasing work to do. This is the one: 551dc838293146345f2c6cdf92fdd20e05f2221f |
The stats subsystem is now capable of shared registrations, so we can simply push the dropped counter registration to LogQueue, even if the parent LogPipe element registers that as well. Signed-off-by: Balazs Scheidler <balazs.scheidler@oneidentity.com>
@bazsi: ok, cherry-picked. |
@bazsi: I think this could to master (please don't forget to approve the PR :) ) |
Build SUCCESS |
No description provided.