[Info-ingres] Waiting indefinitely for the logging system to shut down

[Karl and Betty Schendel]

> On Mar 2, 2020, at 12:59 AM, Steve <s.anderson.au at gmail.com> wrote:
> 
> The other day Ingres would not shutdown.
> 
> Here are the messages in the error log when ingstop was issued:
> 
> ..., lgmisc.c:1582         ]: Tue Feb 18 00:32:19 2020 E_DMA469_P
> ROCESS_HAS_DIED Process (00002F36) has died. A process attached to the logging and locking system has exited without going through n
> ormal cleanup processing. The system will now perform cleanup processing on behalf of the failed process.
> 
> ..., lgkinit.c:1145        ]: Tue Feb 18 00:32:19 2020 E_DMA499_D
> EAD_PROCESS_INFO        Process (00002F36) died with info '(DEFAULT)'.
> 
> 
> Some processes had shutdown, but the name server, archiver and recovery processes were still running.
> 
> When ingstop was issued for the second time, the following message was displayed:
> 
> "Waiting indefinitely for the logging system to shut down"
> 
> Trying to understand what the above means. For example, could the logging system have been waiting on the already shutdown DBMS servers? 


There are a couple possibilities, but in this case the most likely situation is that
because the DBMS server died, the recovery server tried to clean up but
got hung on a stuck spin-lock.  This is an issue that we're aware of and
have on the list to look into at some point.

The other possibility is that the recovery server is actively cleaning up
open transactions, but one or more transactions are large enough that it
is taking a long time to complete the clean-up.

Karl