cancel
Showing results for 
Search instead for 
Did you mean: 

Streaming Replication broke and arch_logs and log directories are growing rapidly.

Level 3 Traveller

Streaming Replication broke and arch_logs and log directories are growing rapidly.

Hi Team,

 

i noticed that the streaming wan broked between mastaer and stadby. When EFM had tried to switchover , but it got failed , with " recovery suspended on failed server node".

 

i also tried to start the EFM cluster but didnot help. Found the below messages from EFM logs:

 


18/01/19 07:16:21 com.enterprisedb.efm.Environment readTriggerFileLocation SEVERE: Could not read the trigger file location from recovery.conf in directory specified in properties file.

 

Please help me on!

2 REPLIES 2
Highlighted
EDB Team Member

Re: Streaming Replication broke and arch_logs and log directories are growing rapidly.

Hi @kiwi,

 

From your query subject and description, there are two issues which you are facing :

 

1) EFM switchover failed to promote the standby server to Master

2) arch_logs and log directories are growing rapidly 

 

Regarding the first issue, from the error it looks like, the location of the trigger file mentioned in the EFM.properties file is not valid or the user do not have access to it.

 

For further analysis, could you please share the below :

 

1) efm.properties files from all the nodes

2) EFM logs from all the nodes

3) Database logs from Master and Standby servers

 

Regarding the second issue, I am assuming the arch_logs is your archive location set in archive_command, for that you can compress the archives from that location and put it in backup location.

 

For the logs issue, we need check the logging parameters, please share the postgrsql.conf file 

 

Please upload all the requested files at https://enterprisedb.sharefile.com/r-rd9137b76d7a4be6a

 

Please get back to us in case of any issues/queries.

 

Regards,

Sudhir

EDB Team Member

Re: Streaming Replication broke and arch_logs and log directories are growing rapidly.

 


@kiwi wrote:

Hi Team,

 

i noticed that the streaming wan broked between mastaer and stadby. When EFM had tried to switchover , but it got failed , with " recovery suspended on failed server node".

That text comes from the recovery.conf file that EFM creates on a failed master node so that it can't be restarted accidentally as a second master db in the cluster. You must have had a failure at some point to get that to happen. Maybe you had a failure, didn't rebuild the failed master db, and are trying to switchover to it anyway? That won't work.

 


i also tried to start the EFM cluster but didnot help. Found the below messages from EFM logs:


18/01/19 07:16:21 com.enterprisedb.efm.Environment readTriggerFileLocation SEVERE: Could not read the trigger file location from recovery.conf in directory specified in properties file.

 


 

After the above, did you stop EFM and then start it? The error above means what it says -- there is no trigger file location in your recovery.conf file, probably because it's the "recovery suspended" file that was generated during a failure. I think you need to fix the old failed master db.

 

For anything else, we would need log files, a description of what happened, etc, as Sudhir wrote.

 

Thanks,

Bobby