[Spread-users] Spread message delays

[John Schultz]

No, I believe in Spread that a daemon is only allowed to send new messages while it holds the token.  I don't think RELIABLE or even UNRELIABLE messages escape that treatment.

So, if the token is lost, then you will pay the Hurry_timeout "cost" before it is recovered.

Cheers!

-----
John Lane Schultz
Spread Concepts LLC
Phn: 301 830 8100
Cell: 443 838 2200

On Mar 28, 2012, at 10:34 AM, Erez Theodorou wrote:

Thanks! I am using a 65b machine! I will give your fix a try!

Also, I was able to recreate the delays on demand, when I introduce packet loss to even a single machine on the network.
I then changed the message types to RELIABLE_MESS (instead of the SAFE_MESS that I had before). According to the spread docs (section 1.3.2), only the SAFE messages need to be received on all spread daemons before they can be delivered. But now I am still seeing the same delay to all the receives even when messages are sent from one node to another on the same machine.

Any ideas?

Thanks!
--Erez

-----Original Message-----
From: John Schultz [mailto:jschultz at spreadconcepts.com] 
Sent: Wednesday, March 28, 2012 3:11 PM
To: Erez Theodorou
Cc: spread-users at lists.spread.org
Subject: Re: [Spread-users] Spread message delays

If you can't fix the loss in your network, then you can lower the delay significantly by lowering the Hurry_timeout in membership.c where it is set around lines 169 and 180.

On a LAN, you can typically safely lower it down to the order of 100's of ms.  On a WAN, depending on how just how wide your WAN is you probably don't want to go below half a second or a second.

Cheers!

-----
John Lane Schultz
Spread Concepts LLC
Phn: 301 830 8100
Cell: 443 838 2200

On Mar 28, 2012, at 8:15 AM, Erez Theodorou wrote:

Hi,

Sometimes messages are only received at the recipient after 1 or 2 seconds.
I have a cluster of nodes and what I see is that all the nodes on the cluster stop receiving messages for the exact same period. Even nodes that send messages to other nodes (processes) on the same machine.
Any ideas what may be the possible causes?

Thanks,
-Erez

_______________________________________________
Spread-users mailing list
Spread-users at lists.spread.org
http://lists.spread.org/mailman/listinfo/spread-users


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3805 bytes
Desc: not available
Url : http://lists.spread.org/pipermail/spread-users/attachments/20120328/71c6f11e/attachment.bin