-
Notifications
You must be signed in to change notification settings - Fork 1.6k
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
0.13.2 crash at msgpack_sbuffer_write #611
Comments
With
How can I figure out which file is being processed when the error occurs? |
would you please check in the log (the content above the one provided) which log file was being processed ? |
Those logs are gone now. I will reopen this issue if I see it again. |
Happened again now :) What is the message I'm looking for that identifies the file being processed? Below I've copied everything form the second before the crash, but omitted lots of "Could not merge JSON as requested" (we do have some not so strict json formatted logs).
|
@edsiper I should rephrase the question actually: If fluent-bit crashes, how do I find out which log file it was processing? If that can be done without switching to debug level, for example through the state db, it would be great. |
We have the same problem. |
reproducer: /tmp/x/filename.log
|
@chlunde FYI - I'm on 0.14-dev and I can't reproduce it there. Using your config and the two lines you gave for reproduction and
|
@michiel I guess it depends on compilation options, OS version and more. This was on RHEL 7.5. When I compiled it with AFL it did not crash on that exact output. Try running it with valgrind, or print out the type after the call to |
@edsiper Will you be pushing 0.14-dev builds to |
@solsson I'm pushing master builds + my PRs to https://hub.docker.com/r/sddmelb/fluent-bit/tags/ The tag |
thanks everyone, the fix provided by @chlunde should fix the problem. I've merged the changes into GIT master and I will release a new 0.13.3 today. |
Unfortunately I get such a crash still
I've tried to switch to the |
I'm having the same issue. I have narrowed it a bit. I have two lines in a log file that causes a SIGSEGV:
If I remove the space between the "now" and "\n" on the second line I don't get the crash. If I remove the first line (leaving in the space in the other line) it doesn't crash but there is no output from the first line. Lines added to the log afterwards are output though. Also if I rename the log file I get a warning 'invalid pattern for given tag kube.var.log.containers....' but it also doesn't crash with the space in it. Hope this helps. |
Signed-off-by: Eduardo Silva <[email protected]>
thanks all for the information and test cases. I've pushed a fix into GIT master which avoids the problem described by @ae-dg . If anyone of you can build from master and test it would be awesome. |
@michiel can you push a new build from master with your travis setup? |
@solsson I've rebased against master and pushed out the changes. As above, this is master + open PR #618 (which only affects Travis build is https://travis-ci.org/michiel/fluent-bit/builds/391740777 - that's now up as tag travis-107 at https://hub.docker.com/r/sddmelb/fluent-bit/tags/ |
I've gone from ~300 crashes per day to zero with the latest fix. Thanks a lot everyone for the effort. |
Signed-off-by: Eduardo Silva <[email protected]>
thanks again, I will do a new release shortly. |
the doc has a duplicated options tag_prefix Signed-off-by: Evgeniy Shubin <[email protected]>
Looks a bit similar to #557 (comment) but that one is reportedly fixed. I have a crashlooping pod with
I'll try to get back with more details as comments.
The text was updated successfully, but these errors were encountered: