[MonetDB-users] mserver5 process locked unkillable
I was testing some queries, and for example this query has made the mserver5 process 100% CPU, which i could not kill and machine could not be rebooted. running Fedora 8 64 bit. query like this: select col1, col2, sum(col3), sum(col4), sum(col5) from thetable group by col1, col2; there are about 160 columns in the table. col1 and col2 are "smallint" col3, col4, col5 are "real" number of distinct values: in col1 is about 120 in col2 about 80 total number of rows in table is about 400M I tested the query 2 times, simpler queries work ok, like this: select col1, count(*) from thetable group by col1; this query has crashed mserver5 before, not sure if this already fixed: select col1, count(1) from thetable group col1; -- View this message in context: http://www.nabble.com/mserver5-process-locked-unkillable-tp16789362p16789362... Sent from the monetdb-users mailing list archive at Nabble.com.
On Sat, Apr 19, 2008 at 06:05:04PM -0700, mobigital1 wrote:
I was testing some queries, and for example this query has made the mserver5 process 100% CPU, which i could not kill and machine could not be rebooted.
running Fedora 8 64 bit.
query like this:
select col1, col2, sum(col3), sum(col4), sum(col5) from thetable group by col1, col2;
there are about 160 columns in the table.
col1 and col2 are "smallint" col3, col4, col5 are "real"
number of distinct values: in col1 is about 120 in col2 about 80
total number of rows in table is about 400M
I tested the query 2 times, simpler queries work ok, like this: select col1, count(*) from thetable group by col1;
this query has crashed mserver5 before, not sure if this already fixed: select col1, count(1) from thetable group col1;
Could you include the table definitions (ie the create table statements), without this its hard to debug. We recently run the SF-100 of tpch which includes a similar query (01) over 600M rows lineitem table, which didn't show problems. So we really need more data to fix this problem. Niels
-- View this message in context: http://www.nabble.com/mserver5-process-locked-unkillable-tp16789362p16789362... Sent from the monetdb-users mailing list archive at Nabble.com.
------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javao... _______________________________________________ MonetDB-users mailing list MonetDB-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-users
-- Niels Nes, Centre for Mathematics and Computer Science (CWI) Kruislaan 413, 1098 SJ Amsterdam, The Netherlands room C0.02, phone ++31 20 592-4098, fax ++31 20 592-4312 url: http://www.cwi.nl/~niels e-mail: Niels.Nes@cwi.nl
this should be the same: table columns: col1..col16 : smallint null col17..col160: real null select col1, col2, sum(col17), sum(col18), sum(col19) from thetable group by col1, col2; should return a total of about 8000 rows, each row would be an aggregation from about 50,000 rows from "thetable". -- View this message in context: http://www.nabble.com/mserver5-process-locked-unkillable-tp16789362p16790887... Sent from the monetdb-users mailing list archive at Nabble.com.
On Sun, Apr 20, 2008 at 05:43:36AM -0700, mobigital1 wrote:
this should be the same:
table columns:
col1..col16 : smallint null col17..col160: real null
select col1, col2, sum(col17), sum(col18), sum(col19) from thetable group by col1, col2;
should return a total of about 8000 rows, each row would be an aggregation from about 50,000 rows from "thetable".
I repeated this experiment using a script to generate random small numbers etc. The loading took about 12 hours (on my desktop (8GB memory)). The query about 13 minutes. I didn't find any problem. Niels
-- View this message in context: http://www.nabble.com/mserver5-process-locked-unkillable-tp16789362p16790887... Sent from the monetdb-users mailing list archive at Nabble.com.
------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javao... _______________________________________________ MonetDB-users mailing list MonetDB-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-users
-- Niels Nes, Centre for Mathematics and Computer Science (CWI) Kruislaan 413, 1098 SJ Amsterdam, The Netherlands room C0.02, phone ++31 20 592-4098, fax ++31 20 592-4312 url: http://www.cwi.nl/~niels e-mail: Niels.Nes@cwi.nl
participants (2)
-
mobigital1 -
Niels Nes