Details
-
Bug
-
Resolution: Fixed
-
Major
-
None
-
None
-
None
-
1
-
SDK30, SDK32
Description
When network link is broken (pf or iptables), libcouchbase will continue retying operations until timeout, but the buffers that has been allocated during operation retry are not getting properly deallocated when the server purges pending operations.
Libcouchbase should clean the IO context object properly in case of network error.
To reproduce, just run some workload in the loop and cut the network link using firewall.
Useful commands for pf on macOS
Block traffic:
$ cat ~/block.conf
|
block drop in from 192.168.107.128
|
block drop in from 192.168.107.129
|
block drop in from 192.168.107.130
|
sudo pfctl -f ~/block.conf && sudo pfctl -e
|
Restore traffic:
sudo pfctl -d && sudo pfctl -F all
|
Attachments
Issue Links
- relates to
-
MB-60934 [Eventing] Fatal error: libcouchbase experienced an unrecoverable error and terminates the program eventing-consumer to avoid undefined behavior
- Open