Re: [Sequoiadb-discuss] Sequoia gets stuck (probably) because of SELECT...FOR UPDATE

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Hi,

 From your log it looks like everyone is waiting for that transaction 65 
to either commit or rollback.
If your thread has crashed and there is no rollback or connection 
closing in a catch block, all the locks are going to be retained forever 
(thought there is a possibility to set timeout on transactions in the 
load balancer settings if I recall).
Without the log of the exact requests, it's hard to tell you more than that.

Keep us posted with your findings,
Emmanuel

On 11/23/2011 06:30, rohit dhamane wrote:
> Hi,
>
> I have been facing some problem with sequoia. I am trying to run 
> BenchmarkSql-2.3 (http://sourceforge.net/projects/benchmarksql/) with 
> sequoia over two postgresql-7.3 servers but it seems to get stuck 
> indefinitely at the same point. I think it's happening because of the 
> SELECT...FOR UPDATE but not really sure how :( It seems that the 
> transactions are getting into a deadlock. Here's the out put when I 
> turned on the debug mode.
>
> 02:56:49,388 DEBUG controller.core.PingResponder Received ping data 
> from /127.0.0.1 <http://127.0.0.1>
> 02:56:49,388 DEBUG controller.core.PingResponder Response to ping sent 
> to /127.0.0.1 <http://127.0.0.1>
> 02:56:49,775 DEBUG sequoia.controller.loadbalancer Building wait-for 
> graph...
> 02:56:49,776 DEBUG sequoia.controller.loadbalancer new_order locked by 64
> 02:56:49,777 DEBUG sequoia.controller.loadbalancer district locked by 65
> 02:56:49,777 DEBUG sequoia.controller.loadbalancer warehouse locked by 65
> 02:56:49,777 DEBUG sequoia.controller.loadbalancer Walking wait-for 
> graph...
> 02:56:49,778 DEBUG sequoia.controller.loadbalancer 64 waits for 
> district locked by 65
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer 68 waits for 
> new_order locked by 64
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer     64 waits for 
> district locked by 65
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer 67 waits for 
> warehouse locked by 65
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer 66 waits for 
> new_order locked by 64
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer     64 waits for 
> district locked by 65
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer Choosing victim node...
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer 0 cycles detected
> 02:56:49,779 DEBUG sequoia.controller.loadbalancer No victim
> 02:56:50,389 DEBUG controller.core.PingResponder Received ping data 
> from /127.0.0.1 <http://127.0.0.1>
> 02:56:50,389 DEBUG controller.core.PingResponder Response to ping sent 
> to /127.0.0.1 <http://127.0.0.1>
>
>
>
> Here's my configuration file..
>
> <?xml version="1.0" encoding="UTF-8"?>^M
> <!DOCTYPE SEQUOIA PUBLIC "-//Continuent//DTD SEQUOIA @VERSION@//EN" 
> "http://sequoia.continuent.org/dtds/sequoia-@VERSION@.dtd">^M
> ^M
> <SEQUOIA>^M
> ^M
> <VirtualDatabase name="tpcw">^M
> ^M
> <AuthenticationManager>^M
> <Admin>^M
> <User username="postgres" password=""/>^M
> </Admin> ^M
> <VirtualUsers>^M
> <VirtualLogin vLogin="postgres" vPassword=""/>^M
> </VirtualUsers>^M
> </AuthenticationManager>^M
> ^M
> <DatabaseBackend name="localhost" driver="org.postgresql.Driver" 
> driverPath="/home/rohit/Software/CJDBC/sequoia-2.10.10-bin/drivers/postgresql7.3.jar" 
> url="jdbc:postgresql://localhost:6551/tpcw" 
> connectionTestStatement="select now()">^M
>
> <ConnectionManager vLogin="postgres" rLogin="postgres" rPassword="">^M
> <VariablePoolConnectionManager initPoolSize="20" minPoolSize="5" 
> maxPoolSize="0" idleTimeout="180" waitTimeout="0"/>^M
> </ConnectionManager>^M
> </DatabaseBackend>^M
> ^M
> <DatabaseBackend name="Remote" driver="org.postgresql.Driver" 
> driverPath="/home/rohit/Software/CJDBC/sequoia-2.10.10-bin/drivers/postgresql7.3.jar" 
> url="jdbc:postgresql://192.168.164.136:6552/tpcw 
> <http://192.168.164.136:6552/tpcw>" connectionTestStatement="select 
> now()">^M
> <ConnectionManager vLogin="postgres" rLogin="postgres" rPassword="">^M
> <VariablePoolConnectionManager initPoolSize="20" minPoolSize="5" 
> maxPoolSize="0" idleTimeout="180" waitTimeout="0"/>^M
> </ConnectionManager>^M
> </DatabaseBackend>^M
> <RequestManager>^M
> <RequestScheduler>^M
> <RAIDb-1Scheduler level="passThrough"/>^M
> </RequestScheduler>^M
> ^M
> <RequestCache>^M
> <MetadataCache/>^M
> <ParsingCache/>                 ^M
> <ResultCache granularity="table" />^M
> </RequestCache>^M
> ^M
> <LoadBalancer transactionIsolation="serializable">^M
> <RAIDb-1>^M
> <WaitForCompletion policy="all"
>             deadlockTimeoutInMs="30000"/>^M
> <RAIDb-1-RoundRobin/>^M
> </RAIDb-1>^M
> </LoadBalancer>^M
> </RequestManager>^M
> ^M
> </VirtualDatabase>^M
> ^M
> </SEQUOIA>^M
>
> If anybody has anu idea why this happens please let me know. I would 
> appreciate it a lot. Thanks in advance.
>
> Regards,
> Rohit
>
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure
> contains a definitive record of customers, application performance,
> security threats, fraudulent activity, and more. Splunk takes this
> data and makes sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-novd2d
>
>
> _______________________________________________
> Sequoiadb-discuss mailing list
> Seq...@li...
> https://lists.sourceforge.net/lists/listinfo/sequoiadb-discuss