Currently we default the indexing request timeout for fluentd to 600 seconds. That is about 10 minutes. In certain situations, this timeout is not long enough. If we make this essentially infinite, we'll avoid fluentd pods re-submitting requests unnecessarily.
There should be no concerns about the infinite timeout. If Elasticsearch is not responding, but has the connection open, then resubmitting the request will not help.
This can adversely affect large clusters with lots of indices, where certain maintenance operations can take longer than 10 minutes.
The fix is not in logging-fluentd/images/v3.9.22-1
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.