Hi Doug,

I uploaded the tcptrace results on
http://groups.google.com/group/hypertable-dev/files.

The bigbuf.tar.gz and smallbuf.tar.gz include the tcptrace results on
each node in the case of 40*32768 and 4*32768, respectively.
Because both files exceed over 10MB, I split these files by split command.
You can get the original tar.gz files
  cat bigbuf.tar.gz.a*  > bigbuf.tar.gz

The Hypertable.Master, Hyperspace.Master and  DfsBroker processes were
in the node named by mario01.dq.isl.ntt.co.jp.
The Hypertable.RangeServer and DfsBroker processes were in the other nodes.

>Oh I see.  So the socket SO_SNDBUF, SO_RCVBUF increase was for the datagram 
>socket inside the method
>Comm::create_datagram_receive_socket()?
Yes.

>If that's the case, then I suspect that Hyperspace is getting bombarded with 
>keepalive packets and dropping them
>because it's socket buffer has filled up.
I think so.

>Another thing for you to try is to set the random startup delay factor to 
>something higher than 5 seconds in the Capfile.  Try something like this:
Ok, I will try later.

Regards.

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"Hypertable Development" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/hypertable-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to