[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[LDM #XGM-432924]: question re usage of ldmping for monitoring system health



Mike,

> We're running the MADIS ldm distribution farm behind a load balancer.
> There was a failure mode a couple of weeks ago where one of the servers
> became unavailable due to software failure, but the load balancer
> didn't detect it.

Did the LDM fail or did something else?

> It occurred to me that since we always have lots of
> ldm connections, if the load balancer could detect ldm failure it
> could take the server out of the farm.

As far as I know, the LDM rarely, if ever, fails.

> I asked our network administrator
> about this and he had the following question:
> 
> "The load balancer software supports a binary health check which I
> believe LDM is a binary protocol?

The LDM uses ONC RPC to communicate, which is a binary protocol.

> I would just need some data to send
> and/or expected data returned... in hex."
> 
> Could he emulate ldmping to do this?  If so, can you provide the actual
> ldmping command and return reply?  Hex would be preferable, but I can
> translate from whatever form it's in.

I've never looked at the messages that the RPC layers exchange when 
establishing a connection, so I don't know what information you would need.

I have used telnet(1) to connect to an LDM port (388).  telnet(1) doesn't 
exchange anything during the connection setup other than the normal 3-way TCP 
handshake.  It should suffice, therefore, to simply establish a TCP connection 
to port 388 to determine if an LDM is running on the remote host or not.

> Thanks,
> 
> Mike


Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: XGM-432924
Department: Support LDM
Priority: Normal
Status: Closed