[Bug 1158] pmcd observed to return PMNS_IDS PDU in response to FETCH

[bugzilla-daemon@xxxxxxxxxxx]

Comment # 9 on bug 1158 from Ken McDonell

Indeed.

I'm mulling over a possible change in the implementation that would mean when
using a host context, if __pmGetPDU returns PM_ERR_TIMEOUT *or* PM_ERR_IPC,
then we should close the underlying socket.

All contexts sharing the socket would see an error at the next attempt to send
or receive a PDU.

It would drive all clients into their "exit or pmReconnectContext()" logic, but
would avoid the synchronous protocol between pmcd and a client being
compromised, as I believe is happening here.

Using the debug flag -Dpdu, rather than -Dpdu,fetch would have been better in
this case because the extra PDUs associated with the -Dfetch diagnostics have
probably made a bad situation worse.

---

      You are receiving this mail because:

• You are on the CC list for the bug.
• You are the assignee for the bug.