Re: log file not found after a crash

Just to give some more information: MonetDB 11.9.7 repeatedly crashes on some long running queries. There is no problem with the HDD, because the rest of the programs and OS work fine. For the long running queries, I see the physical memory consumption going on increasingly steadily and then MonetDB crashes by printing message on the command prompt -- MAPI = (monetdb) /tmp/.s.monetdb.5678 ACTION= read_line QUERY = select count(t1.sub||t1.obj||t2.sub) from uniprot800m_p37_obj as t1, uniprot800m_p70_obj as t2 where t1.obj=t2.obj; ERROR = !Connection terminated Tables uniprot800m_p37_obj and uniprot800m_p70_obj are created by ordering on column "obj" first and then on "sub" (so the query is actually a simple merge-join). Also another thing noted was -- when MonetDB starts crashing, it ends up consuming almost entire remaining empty space on the disk (the original datasize is 48GB, but it consumes additional 150+ GB filling up the entire disk). I am wondering if there is something wrong with the way query is written or if there is some other problem.. On Sat, Oct 20, 2012 at 7:31 AM, Medha Atre wrote:

This time it again crashed while running queries in bulk. I had written a script to restart the server after every 50 queries because the last time I had also got a message "too many open files" on the command prompt after the MonetDB crash.

This time merovingian.log under under my dbfarm showed messages like "database 'uniprot' (3469) was killed by signal SIGBUS".

I also checked /var/syslog and saw messages like -- "INFO: task mserver5:4006 blocked for more than 120 seconds.", ""echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message." with timestamp around the time MonetDB had crashed, with a "Call trace:" given after that.

On Sat, Oct 13, 2012 at 10:43 AM, Martin Kersten wrote:

...

Hello

select * from environment()

gives an overview of the locations of things. The merovingian.log should recide in the gdk_dbfarm folder

If your disk/system experienced a hard crash, then both database and log may be broken. Then you have to recover from backup

regars, Martin

On 10/13/12 4:26 PM, Medha Atre wrote:

...

Hi,

I am running MonetDB v. 11.9.7. I have a large RDF dataset loaded in it (800+ million triples) as separate predicate tables for each unique predicate. I am running some simple one join SPAQRL queries ("?s :p1 ?o JOIN ?o :p2 ?x" kind) in bulk through a shell script (echo "<some query>" | mclient -drdf -lsql >> results). Sometime overnight when the queries were running, an error occurred in the database run, and it crashed by spitting out messages

"monetdbd: an internal error has occurred, refer to the logs for details, please try again later" and "monetdbd: internal error while starting mserver, please refer to the logs"

for the rest of the queries after some query caused this error.

But when I tried to open "<Path To MonetDB installation>/var/log/monetdb/merovingian.log", it said "no such file or directory". I restarted the server and the particular database. It shows me its status as "health 97%" and "crash <crash date and time>". I tried to find the logfile again, but it's not there.

I am wondering how I can know what the problem was. _______________________________________________ users-list mailing list users-list@monetdb.org http://mail.monetdb.org/mailman/listinfo/users-list

_______________________________________________ users-list mailing list users-list@monetdb.org http://mail.monetdb.org/mailman/listinfo/users-list

users-list mailing list users-list@monetdb.org http://mail.monetdb.org/mailman/listinfo/users-list