Details
-
Improvement
-
Status: Open
-
Minor
-
Resolution: Unresolved
-
0.10.2.3
-
None
-
None
-
Linux 3.10.0-693.2.2.el7.x86_64
Description
I use kafka 10.0.2 in the production environment, and a problem occurs: the problem should be after the re-election of the controller. due to the relatively large amount of data, the log is soon filled, and soon the alarm of insufficient disk space appears . The log level used in our production environment is "info", and the log is like this:
The ephemeral node [***] at *** has gone away while reading it
I downloaded the source code of kafka from github and found in kafka.utils.ZkUtils
CreateCallback#:
case Code.NODEEXISTS => zkHandle.getData(path, false, getDataCallback, null)
GetDataCallback#:
case Code.NONODE => info("The ephemeral node [%s] at %s has gone away while reading it, ".format(data, path))
one case is Code.NODEEXISTS , but another case is Code.NONODE. I think this problem should come from zookeeper, but I don’t know why?
Please help me, thanks!