Hi, While I was running a query today, I got this error today - *Allocation failedGDKmallocmax: failed for 18446744072964793488 bytes* I ignored it thinking it was some runtime error and repeatedly tried to run the same query 3 or 4 times. Finally, mclient got crashed. I tried to reconnect to mclient, but it was not successful. So I tried restarting the database, only to find that the database has already been shut down. When I tried starting the database, I am receiving this error - *starting database 'testDB'... FAILEDstart: starting 'testDB' failed: database 'testDB' appears to shut itself down after starting, check monetdbd's logfile for possible hints* I tried after restarting the demon multiple times, but no use. I checked merovingian.log file in the db farm folder, but it showed only this - *2015-02-04 05:59:50 ERR control[7829]: (local): failed to fork mserver: database 'testDB' appears to shut itself down after starting, check monetdbd's logfile for possible hints* I tried to check the monetdbd's log file, but I am unable to find the log folder in var. Has the database crashed irrecoverably? How to run MonetDB again with at least a new database?? Any help much appreciated. P.S.: I run MonetDB Oct2014-SP1 version in a CentOS machine with 128 GB RAM. Thanks & Regards, Vijay.
On 04-02-2015 19:46:35 +0530, Vijay Krishna wrote:
starting database 'testDB'... FAILED start: starting 'testDB' failed: database 'testDB' appears to shut itself down after starting, check monetdbd's logfile for possible hints
I tried after restarting the demon multiple times, but no use.
I checked merovingian.log file in the db farm folder, but it showed only this -
2015-02-04 05:59:50 ERR control[7829]: (local): failed to fork mserver: database 'testDB' appears to shut itself down after starting, check monetdbd's logfile for possible hints
I tried to check the monetdbd's log file, but I am unable to find the log folder in var.
It's a bit confusing. merovingian.log actually /is/ monetdbd's logfile. Perhaps in a future release of MonetDB, the default name for the logfile (in new dbfarms) can be monetdbd.log instead. However, if you scroll up in the log you quoted above, you should see something about starting the database 'testDB'. It should also give you more information about how the database terminated. Something along the lines of "TERM", "KILL", "SEGV" is likely to be found there. Fabian
Hi On 04/02/15 15:16, Vijay Krishna wrote:
Hi,
While I was running a query today, I got this error today -
*Allocation failed GDKmallocmax: failed for 18446744072964793488 bytes* The kernel plans to allocate an enormous amount of memory, which evidently fails. It should lead to a normal stop of the system, causing the SQL transaction to fail.
I ignored it thinking it was some runtime error and repeatedly tried to run the same query 3 or 4 times. Finally, mclient got crashed.
Ignoring such messages should not be taken lightly. They indicate either (mostly) a flaw in your query (e.g. asking the system to compute a Cartesian product) or bouncing upon a resource limitation (e.g. running on a 32-bit platform)
I tried to reconnect to mclient, but it was not successful. So I tried restarting the database, only to find that the database has already been shut down.
Be aware that the system might need time to redo the transaction log with all changes that needs to be re-applied. Repeated hard stops in such situations are a potential cause of corruption of the status.
When I tried starting the database, I am receiving this error -
*starting database 'testDB'... FAILED start: starting 'testDB' failed: database 'testDB' appears to shut itself down after starting, check monetdbd's logfile for possible hints*
I tried after restarting the demon multiple times, but no use.
The monetdb deamon does not help here/
I checked merovingian.log file in the db farm folder, but it showed only this - *2015-02-04 05:59:50 ERR control[7829]: (local): failed to fork mserver: database 'testDB' appears to shut itself down after starting, check monetdbd's logfile for possible hints *
A better sanity check during recovering from forced servers termination is an area that calls for some improvement (when time and research agendas permit)
I tried to check the monetdbd's log file, but I am unable to find the log folder in var.
Has the database crashed irrecoverably? How to run MonetDB again with at least a new database??
Most likely the system has been brought into a really corrupted state. A fresh start from back up is what is the next step.
Any help much appreciated.
P.S.: I run MonetDB Oct2014-SP1 version in a CentOS machine with 128 GB RAM.
Thanks & Regards, Vijay.
regards, Martin
_______________________________________________ users-list mailing list users-list@monetdb.org https://www.monetdb.org/mailman/listinfo/users-list
participants (3)
-
Fabian Groffen -
Martin Kersten -
Vijay Krishna