Conference azur::mcc

5572.0. "Alarms causing resource problem/errors" by BLKPUD::MORANT (Wanted: AK47 assault rife and 200 rounds!) Tue Sep 07 1993 07:05

Hi,

5000/240, 48MB, approx 200MB swap space.
Ultrix 4.3, BMS 1.3 (LAN manager 200), DECdts (latest), Local namespace
DECnetIV
Alarms:

1) SNMP GIGA_SWITCH ipReachability = up, at every 00:00:10
2) NODE * state = on, at every 00:00:10
3) NODE4 * state = on, at every 00:00:10
4) NODE LOCAL_NS:.fotmi Routing Circuit circuit-2 state = on, at every 00:00:10
5) Concentrator * Phy Port * Phy Port state = watching, at every 00:00:10
6) Concentrator * Phy Port * Phy Port state = waiting, at every 00:00:10
7) Concentrator * Phy Port * Phy Port state = in use, at every 00:00:10
8) Bridge LOCAL_NS:.bridge520_1 line1 Port module state = forwarding, at every
								00:00:10

My customer (DEC consultant tring to do a demo on site) is experiencing a 
fairly strange problem. His system is as detailed above. He has 8 alarms 
running (detailed above) and every 1.5-2hrs the error messages listed below 
occur. If he's quick, and disables the alarms and then re-enables them 
everything returns to normal for another 1.5-2hrs. However, if he sits back 
and does nothing at this point, the session manager appears to hang, ie he 
can't switch focus to another window or create a DECterm. Sadly, there are no 
other available methods of getting onto this machine (ie no directly or 
otherwise attached terminals or other available machines on the network) so we 
can't tell what is exactly going on, and he has to re-boot to clear the 
problem.

The error messages are (and they seem to realate to the concentrator alarms):

%MCC-F-STRTDEVERROR, Start Ethernet device failed
%MCC-W-TIME_ALREADY_PASSED, Scheduled time has already passed
%MCC-E-PROTOCOL_IN_USE, Protocol in use and not available
%MCC-E-INV_HANDLE, Invalid Handle rule
One or more events were lost in domain
In addition there's a sprinkling of "Software Error" and "Requested operation
cannot be completed".

Can anyone help please? If you need further info, then please ask.

Tim Moran UK/CSC
    

    I would suggest you change the time semantics on the concentrator rules
    to be more like event 1-5 minutes, not every 10 seconds. I know
    you want fairly real time but there have been a number of complaints
    regarding alarming on concentrators with too high a frequency.