This can happen because of networking problems. TCP would help you
with its flow control. Can you give us details about why the channels
with TCP did not work?
cheers,
nuno
On Mar 26, 2007, at 3:58 , Sylvain Coutant wrote:
Hi,
We're quite new to Sequoia but were able to setup a dual controller
conf running through Appia sync without too much trouble.
The setup is dual controller, each running two DB backends (MySQL).
The conf is RAIDb-1. We were unable to have appia communicate with
anything else than "UDP p2p TOKEN Channel". TCP channels did not
connect (we have a strange setup for ethernet controllers,
involving vlans and such).
As long as the system is used with slow rate (a couple per second)
INSERT queries, it works quite well. When we launch 50 threads,
each running 100 INSERT queries, the controllers stops seeing each
other at some point. Log from one controller :
16:39:50,909 DEBUG continuent.hedera.adapters Replying to Member
(address=/192.168.0.27:12000, uid=192.168.0.27:12000) for message 1616
16:39:50,913 INFO controller.virtualdatabase.sequoia Member
(address=/192.168.0.27:12000, uid=192.168.0.27:12000) see members:
[Member(address=/81.19.18.26:12000, uid=81.19.18.26:12000), Member
(address=/192.168.0.27:12000, uid=192.168.0.27:12000)] and has
mapping:{Member(address=/192.168.0.27:12000, uid=192.168.0.27:12000)
=sequoia02.testplatform.com:1090}
16:39:50,918 DEBUG continuent.hedera.gms Appia reported block()
16:40:04,466 DEBUG continuent.hedera.gms Appia reported new view:
version: 1
group: [Group:sequoia]
id: [ViewID:4;[AppiaHedera@/192.168.0.27:12000]]
previous: [[ViewID:3;[AppiaHedera@/81.19.18.26:12000]],]
view: [[AppiaHedera@/192.168.0.27:12000],]
addresses: [/192.168.0.27:12000,]
16:40:04,467 INFO continuent.hedera.gms Member(address=/
81.19.18.26:12000, uid=81.19.18.26:12000) failed in Group(gid=sequoia)
16:40:04,469 INFO controller.virtualdatabase.sequoia 25 requests
were waiting responses from Member(address=/81.19.18.26:12000,
uid=81.19.18.26:12000)
16:40:04,469 DEBUG continuent.hedera.adapters Received reply from
Member(address=/192.168.0.27:12000, uid=192.168.0.27:12000) to
message 1610
16:40:04,470 DEBUG continuent.hedera.adapters Received reply from
Member(address=/192.168.0.27:12000, uid=192.168.0.27:12000) to
message 1611
16:40:04,470 WARN controller.RequestManager.sequoia 1 controller
(s) died during execution of request 562949953422915
16:40:04,471 WARN controller.RequestManager.sequoia Controller
Member(address=/192.168.0.26:12000, uid=192.168.0.26:12000) is
suspected of failure.
16:40:04,471 WARN controller.RequestManager.sequoia 1 controller
(s) died during execution of request 562949953422916
16:40:04,471 WARN controller.RequestManager.sequoia Controller
Member(address=/192.168.0.26:12000, uid=192.168.0.26:12000) is
suspected of failure.
16:40:04,471 WARN controller.RequestManager.sequoia 1 controller
(s) died during execution of request 562949953422917
I must say the sequoia controllers run on small virtual test
systems with very low CPU capabilities.
Any help appreciated to sort this out.
Regards,
Sylvain.
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
--
Nuno Carvalho
University of Lisbon, Portugal
http://dialnp.di.fc.ul.pt
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
