[Spread-users] Is there a way for a send-only client to detect that it's messages were received?

Dave Koberstein davek at mojix.com
Sun Jun 29 01:19:13 EDT 2008


It seems like from your definition that there is only one receiver.  How 
about if the receiver replies to the sender's private group - which is 
basically a direct message back to the sender whether or not that sender 
is subscribed to any groups.

The sender could block waiting for the reply before disconnecting.

One of my groups has essentially an RPC service.  The sender is 
expecting information back in that case, however the pattern is 
basically to send to a group to which it is not subscribed and then 
block waiting for the reply.  It disconnects from spread after receiving 
the reply.

Davek

Allan Bailey wrote:
> I'm considering considering using spreadlogd to receive log messages as
> way to replace our current apache and application log file collection
> mechanism.
>
> I read a previous thread that mentioned that a client's messages will be
> dropped on disconnect if they have not be acknowledged by all nodes in
> the ring.
>
> My test has been this:
>
> 3 node ring.
>
> client script that dumps a 10 million line logfile into a 'test' group.
> Using python, and SpreadModule-1.5.  Connection is with
> spread.RELIABLE_MESS, and no group membership.  Just multicast'ing to
> the 'test' group.
>
> spreadlogd that logs all lines sent to the 'test' group to a file.
>
> At the end, i have 9.7m lines in the spreadlogd file.
>
> My question:  is there some way for the client to detect that it should
> wait for disconnect?  Or is there some convention for client disconnect?
>
> So far from reading the spread and libsp man pages I have found nothing
> that indicates there is a way to poll the ring to see if the messages
> have been delivered.
>
> --
> Allan
>
> _______________________________________________
> Spread-users mailing list
> Spread-users at lists.spread.org
> http://lists.spread.org/mailman/listinfo/spread-users
>   





More information about the Spread-users mailing list