Problem with enterprise replication

Informix V10.FC10 Sun Solaris 5.9 The instance reports Blocked:DDR at the moment. How do I free the instance from this state? The following message appeared in the log file last night: 21:08:37 DDR Log Snooping - DDRBLOCK phase started, userthreads blocked Thanks Andy Grantham.

Andrew Is this on the source or the target ? 1. Is the 'other' IDS instance still running correctly ? 2. Is there an open transaction spanning you entire logical logs ? 3. onstat -g ddr onstat -rqm BRIEF onstat -g dss IDS Version(s), O/S(s), ER Setup ??? Keith On 2 August 2011 08:42, Andrew Grantham <agrantha@hotmail.com> wrote: > Informix V10.FC10 > Sun Solaris 5.9 > > The instance reports Blocked:DDR at the moment. How do I free the instance > from this state? > > The following message appeared in the log file last night: > > 21:08:37 DDR Log Snooping - DDRBLOCK phase started, userthreads blocked > > Thanks > > Andy Grantham. > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > >

Keith, Informix version and O/S are in the original email. ER setup is 11 source instances on two seperate database servers replicating to 2 databases in 1 instance on a third server. 2 databases because they use different locales. All replicates are master/slave. The target instance was online and reporting no problems. No open transactions I could see. The instance has been running for a month without problems although I have had to add an extra chunk to the special space. I cannot give you any onstat commands as replication on the instance has been dropped ( following orders ). Regards Andy G > To: ids@iiug.org > From: smiley73@gmail.com > Subject: Re: Problem with enterprise replication [24488] > Date: Tue, 2 Aug 2011 05:56:20 -0400 > > Andrew > > Is this on the source or the target ? > 1. Is the 'other' IDS instance still running correctly ? > 2. Is there an open transaction spanning you entire logical logs ? > 3. onstat -g ddr > > onstat -rqm BRIEF> > onstat -g dss> > IDS Version(s), O/S(s), ER Setup ??? > > Keith > > On 2 August 2011 08:42, Andrew Grantham <agrantha@hotmail.com> wrote: > > Informix V10.FC10 > > Sun Solaris 5.9 > > > > The instance reports Blocked:DDR at the moment. How do I free the instance > > from this state? > > > > The following message appeared in the log file last night: > > > > 21:08:37 DDR Log Snooping - DDRBLOCK phase started, userthreads blocked > > > > Thanks > > > > Andy Grantham. > > > > > > > ******************************************************************************* > > Forum Note: Use "Reply" to post a response in the discussion forum. > > > > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. >

This will generally happen because you are spooling and the spool sblob is full. If that is the case then you need to enlarge the sblob. Also you can add a new log file. From: "Andrew Grantham" <agrantha@hotmail.com> To: ids@iiug.org Date: 08/02/2011 02:45 AM Subject: Problem with enterprise replication [24487] Sent by: ids-bounces@iiug.org Informix V10.FC10 Sun Solaris 5.9 The instance reports Blocked:DDR at the moment. How do I free the instance from this state? The following message appeared in the log file last night: 21:08:37 DDR Log Snooping - DDRBLOCK phase started, userthreads blocked Thanks Andy Grantham. ******************************************************************************* Forum Note: Use "Reply" to post a response in the discussion forum.

Andy Sorry, read the 'important' bit and missed the 'some time very important, but often information only' bit :-)) DDRBLOCK only occurs when the Replay is unable to move (or only moves slowly) and gets caught by the Current. This can be because the target is not responding correctly after applying transactions or (more often I have found) a long transaction on the source, either because someone hasn't Committed or RolledBack or is updating umpteen million rows in a single transaction. Bouncing the target can often shake things loose but if you've has to trash ER then there is not much diagnosis possible. I have a small job running every 15 minutes on my source which alarms if the replay gets more than 3 logs (out of 30) behind the current position. Some false alarms but has saved the instance many times !! due to raw SQL sessions not closing transactions. Keith On 2 August 2011 13:58, Andrew Grantham <agrantha@hotmail.com> wrote: > Keith, > > Informix version and O/S are in the original email. ER setup is 11 source > instances on two seperate database servers replicating to 2 databases in 1 > instance on a third server. 2 databases because they use different locales. > All replicates are master/slave. The target instance was online and reporting > no problems. No open transactions I could see. The instance has been running > for a month without problems although I have had to add an extra chunk to the > special space. I cannot give you any onstat commands as replication on the > instance has been dropped ( following orders ). > > Regards > > Andy G > >> To: ids@iiug.org >> From: smiley73@gmail.com >> Subject: Re: Problem with enterprise replication [24488] >> Date: Tue, 2 Aug 2011 05:56:20 -0400 >> >> Andrew >> >> Is this on the source or the target ? >> 1. Is the 'other' IDS instance still running correctly ? >> 2. Is there an open transaction spanning you entire logical logs ? >> 3. onstat -g ddr >> >> onstat -rqm BRIEF>> >> onstat -g dss>> >> IDS Version(s), O/S(s), ER Setup ??? >> >> Keith >> >> On 2 August 2011 08:42, Andrew Grantham <agrantha@hotmail.com> wrote: >> > Informix V10.FC10 >> > Sun Solaris 5.9 >> > >> > The instance reports Blocked:DDR at the moment. How do I free the instance >> > from this state? >> > >> > The following message appeared in the log file last night: >> > >> > 21:08:37 DDR Log Snooping - DDRBLOCK phase started, userthreads blocked >> > >> > Thanks >> > >> > Andy Grantham. >> > >> > >> > >> > ******************************************************************************* >> > Forum Note: Use "Reply" to post a response in the discussion forum. >> > >> > >> >> >> > ******************************************************************************* >> Forum Note: Use "Reply" to post a response in the discussion forum. >> > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > >

Madison, I did add another chunks to the sblob space on the 26th July but I only got the problem this morning. I did check the free space at the time and it did not suggest this was a problem. I also added more logical log files and this did not solve the problem either. Unfortunately I had to kill replication for this instance as there were locks that prevented other applications running so diagnosis is going to be hard now. The instance itself is very small and the likelyhood of large transactions is slim to say the least. It has been working perfectly well for a good month or so without problems other than requiring more space to the sblob. Should the sblob need more space or was this symptomatic of a problem? If so what is worth looking at now as I need to re-enable replication for this instance and may as take any remedial steps whilst doing that. Also the delete server statement did not completely remove the instance from replication. I can see on the master server that the instance is marked as disconnected and on the instance itself the syscdr db has not been removed. Can I issue delete server statements locally to clean up the environment or is another way better? Regards Andy G. > To: ids@iiug.org > From: mpruet@us.ibm.com > Subject: Re: Problem with enterprise replication [24492] > Date: Tue, 2 Aug 2011 09:08:35 -0400 > > This will generally happen because you are spooling and the spool sblob is > full. If that is the case then you need to enlarge the sblob. > > Also you can add a new log file. > > From: "Andrew Grantham" <agrantha@hotmail.com> > > To: ids@iiug.org > > Date: 08/02/2011 02:45 AM > > Subject: Problem with enterprise replication [24487] > > Sent by: ids-bounces@iiug.org > > Informix V10.FC10 > Sun Solaris 5.9 > > The instance reports Blocked:DDR at the moment. How do I free the instance > from this state? > > The following message appeared in the log file last night: > > 21:08:37 DDR Log Snooping - DDRBLOCK phase started, userthreads blocked > > Thanks > > Andy Grantham. > > > ******************************************************************************* > > Forum Note: Use "Reply" to post a response in the discussion forum. > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. >