OK I've done google, searched forums and faqs. Found several mentions of the 139 out of bounds but no answers that seem relevant.
We have Nagios running on FreeBSD. It monitors multiple Unix/Linux/FreeBSD clients that run nrpe. It also monitors multiple Windows clients that run nsclient.
On one of the Windows clients which has been giving reasonable statuses since it was setup for monitoring in July we are now seeing:
return code of 139 is out of bounds
for all 4 of the defined services.
The defined services are all using the check_nt command.
Doing a test from command line:
check_nt -H fakeservername -p 1248 -v SERVICESTATE -l "Norton AntiVirus Server"
Segmentation fault (core dumped)
However there is no core file left anywhere on the Nagios host we run this from. Since the same command above works fine for all other Windows servers I don't believe it is an issue with the check_nt binary.
The Windows admins removed and reinstalled nsclient on the affected host with no improvment.
Some of the discussion I've seen talks about nslookup issues but I've verified nslookup from the Nagios master host can resolve the server by both its short name and its FQDN so that isn't an issue.
One post I saw had a suggestion by the user that this was caused by FreeBSD protecting itself against a DOS attack but it never gave a final answer as to whether that was the case. It hardly seems likely since we're only getting this issue from one host.
I can telnet to the host on port 1248 so it is listening there.
Does anyone have any ideas?