-
Notifications
You must be signed in to change notification settings - Fork 876
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
Falco took so long to write a JSON output log to stdout #1198
Comments
Not sure, but... might it be related to the throttling configuration? |
It could be related. I did some more experiments and found that the outputs seem the get buffered although the So now the question becomes that why the outputs get buffered although the configuration is set to |
|
Any updated about this bug?
This value I have been trying to find where buffered_output bug might be but I am not very familiar with the codebase so haven't found anything yet. Another possible place where falco/userspace/falco/falco.cpp Line 453 in 3bfd94f
|
Is there any config that will make the buffered output flush faster? at least to workaround the issue. |
I have also tried:
But the problem persists. Since it happens only when Falco runs inside a container, that lets me think that the problem persists even without the buffered output. We need some time to dig more into it. |
If it helps, if I start the container as an interactive TTY like the following but If I start it as a daemon like this In this last example I generate events like spawning a shell into a container, and all the logs show up at the same time after I send ~10 to ~15 events. The weird part is that no log is generated if I generate just 1 event I waited up to an hour and didn't show in the logs. |
Any news on this? I have deployed it with apt in me regular vms and baremetal, but I want to deploy to GKE and it's an issue as it need to run in a container there. |
Hi @fcastello and @smeeklai I believe I found the root cause of this problem, PTAL 👉 #1296 |
Hey @fcastello and @smeeklai Since #1296 got merged, you can try it using the following docker image Let me know if that solved your problem! Thanks in advance. |
@leogr I can't test it yet as the master image has errors. So I will need to wait until someone else fixed that in master image |
My fault, I will fix it shortly.
…On Tue, Jul 7, 2020, 21:45 Francisco Castello ***@***.***> wrote:
@leogr <https://github.com/leogr> I can't test it yet as the master image
has errors.
I am getting this when it starts
/docker-entrypoint.sh: line 20: syntax error in conditional expression:
unexpected token ;' /docker-entrypoint.sh: line 20: syntax error near ;'
/docker-entrypoint.sh: line 20: if [[ ! -z "${SKIP_MODULE_LOAD}"]]; then'
So I will need to wait until someone else fixed that in master image
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#1198 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAZ34FM362FGNRY3PDDYLIDR2N3MLANCNFSM4M3G7XCA>
.
|
Hey @fcastello I have fixed that problem by #1307 (already merged into master). And sorry for the inconvenience 👼 |
Thank you @leogr I already pulled the image and started using it. The original issue is fixed for me, thank you very much! |
/close
On Fri, 10 Jul 2020 at 23:50, Francisco Castello ***@***.***> wrote:
Thank you @leogr <https://github.com/leogr> I already pulled the image
and started using it. The original issue is fixed for me, thank you very
much!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#1198 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAA5J4ZLNU6YLMCCR6MSJETR26EI3ANCNFSM4M3G7XCA>
.
--
L.
|
Describe the bug
I know that Falco would take sometime to evaluate and output a log when a certain condition met. For example, I tried to
exec
into a container and found that Falco took 2-3 mins to output that log tostdout
. However, after I've added several custom rules, now Falco took so long to output the logs, or sometime doesn't output the logs at all until I or K8S master delete the pods. I also found that the timestamp that's in the log is actually not so far from when the action was performed. But the time until that log was output tostdout
is surprisingly long. You can see in the screenshot, thetimestamp
represents the time which is taken from Falco (since it's the same), and thereceiveTimestamp
which represents the timeFluentd
received the log. The different is about ~10 mins.How to reproduce it
Expected behaviour
The time it take the logs to output should be about the same or not much different compared to when I haven't added any custom rules.
Screenshots
Environment
GKE version
1.14.10-gke.27
Falco version:
0.22.1
System info:
GKE
COS
4.14.138+
Additional context
Container runtime version:
docker://18.9.7
The text was updated successfully, but these errors were encountered: