Overview
You notice that all AppleShare users were disconnected from a Xinet shared volume at the same time. When reviewing the at_log, the following error messages are seen:
- ksd(): Errno 32 on TCP send to <IP address>
- ksd(): TCP Session "<IP address>" timed out (Client crashed?)
- ksd(): TCP Session <IP address> closed - 50 bad receives (Client crashed?)
- ksd(): TCP Session <IP address> closed - read error 110 (Client crashed?)
- ksd(): TCP Session <IP address> closed - read error 104 (Client crashed?)
Solution
This behavior is most often related to a networking issue, however, there are a number of reasons why a connection will reset in this way. For example, switching between network interfaces, DHCP reconfigurations, rebooting a firewall, general packet corruption, misrouting, etc.
When seeing these errors, it can most often be traced to rogue/orphaned ksd
processes that have failed to close.
- Stop the Xinet Services:
# /usr/etc/appletalk/atinit halt
- Search for rogue
ksd
services still running:
# ps -edf | grep ksd
- Example Output:
-
[root@localhost xinet]# ps -edf | grep ksd
root 2862 1 0 Apr16 ? 00:00:00 /usr/etc/appletalk/ksd -f -v -o -i -longcomm ip-address
root 2870 2862 0 Apr16 ? 00:00:00 /usr/etc/appletalk/ksd -f -v -o -i -longcomm ip-address
wnv 2871 2862 0 Apr16 ? 00:00:05 /usr/etc/appletalk/ksd -f -v -o -i -longcomm ip-address
root 10803 2862 0 Apr18 ? 00:00:00 /usr/etc/appletalk/ksd -f -v -o -i -longcomm ip-address
root 10977 2862 0 Apr18 ? 00:00:01 /usr/etc/appletalk/ksd -f -v -o -i -longcomm ip-address
root 21946 21936 0 12:31 pts/0 00:00:00 grep --color=auto ksd - Note: The main KSD process highlighted (with PID 2862 on the output above) would be the parent of all other spawned KSD processes.
-
- Example Output:
- Once located, you can use the "kill" command on the rogue
ksd
processes PID to properly unregister it:# kill <PID>
- Restart the Xinet Services:
# /usr/etc/appletalk/atinit start
Priyanka Bhotika
Comments