-
Notifications
You must be signed in to change notification settings - Fork 578
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
Icing2 2.11.2-1 HA Master sync overhead #7711
Comments
Likely related to #7687 @lippserd @Al2Klimov |
Well, each node queues its cluster messages in RAM and caches them on the HDD. The more messages/time, the more traffic and I/O happens. And if your resources aren't enough for that traffic and I/O, your RAM and HDD usage grows (of course). It's kinda the same as you get "your DB isn't able too keep up" form IDO in too large setups. IMAO it's not an actual memory leak nor related to the crash. |
I didn't saw any disk usage or caching on HDD at icinga2 masters during all tests(replay log disabled), only RAM usage increase. |
By the way: On OOM Linux kills the waster, but the JSON-RPC crash is about a SEGV. |
@lippserd observed a memory leak within the replay log handling, which is seemingly gone with the git master (and highly likely fixed with the JSON library update). This could play a role here. |
I'll be able to provide you additional test results once 2.12 / 2.12RC will be released |
It's been released, please test it. |
Hello @Al2Klimov, yeah i already saw the blog post, great news and thanks for all the work that have been done. |
IMAO the lack of external feedback for a long time indicates that that feedback will never happen. Therefore closing this one. Feel free to re-open if the problem persists with the latest Icinga 2 version as long as you provide the desired information. |
Describe the bug
Icinga2 in Master in HA mode 3 times slower compared to Icinga2 Master in standalone mode
On high load i face memory leaks in Icinga2 Master HA setup, on same or bigger workloads memory leaks are not reproduced in Icinga2 master in standalone mode
To Reproduce
Provide a link to a live example, or an unambiguous set of steps to reproduce this bug. Include configuration, logs, etc. to reproduce, if relevant.
Test results with comparison and attached graphs/logs described in https://community.icinga.com/t/icinga2-at-large-scale/2178, comparison between Test 3 and Test 5 the most important.
Expected behavior
Expecting nearly the same performance of Icinga2 master in HA as i have with Standalone mode
Log comparison
"ApiListener, RelayQueue" related logs
HA mode, 300000 services, 1 min frequency:
Standalone mode, 500000 services, 1 min frequency:
Your Environment
Include as many relevant details about the environment you experienced the problem in
icinga2 feature list
The text was updated successfully, but these errors were encountered: