- users-list - monetdb.org

RE: Details about MonetDB, data loss in restart server (Ying Zhang)
by Luciano Sasso 08 Apr '15

08 Apr '15

Hi, - OS is Debian 7 64-bit, in a VM in VMware virtualization System. - Hard disk 100G with 26% usage. - 5 cores. - 6G RAM. - Version MonetDB: MonetDB 5 server v11.15.17 "Feb2013-SP5" - Table: create table (id int, dt date, product varchar (50) .... etc, and another 30 columns between int and varchar. - Query: select dt, count (id) from table where produtct in (....) group by dt - History: Long time MonetDB use with Pentaho (Saiku) for analysis of data cubes, never had problems, works great with millions of records. When the problem occurred, the steps taken were: 1- A couple of days before it occurred, I made a 32 million load records to the table. 2- I used this table for many analyzes via SQL client (without using cubes) and everything was fine. 3- I created a cube in Pentaho for this table and also used a lot and everything went well. 4- When I put the dt fields in the cube and the interface seems to have caught the Pentaho and mserver held all machine resources (memory and CPU), less disk. This lock may have been because the Pentaho, but specifically the mondrian, have had a problem with the date field 5- I tried to stop the Pentaho, but did not work, had to force a kill -9 pid. 6- In the case of MonetDB, the command MonetDB stop dw, worked properly and stopped the service. 7- When I run MonetDB start dw, it took a while longer than usual, but the service start usually without showing error on console. 8- With other tables there were no data loss problem, only that table lost data just the product column, which was used in the query, the remaining columns have not suffered problems. But but at no time update or delete command has been executed, only selects were executed 9- All though I forced Pentaho standing in the execution of the query, but MonetDB stopped normal, and a query should never change data, of course? Log 2015-03-31 15:58:38 MSG control[1114]: (local): served status list 2015-03-31 15:58:38 MSG merovingian[1114]: sending process 14173 (database 'dw-zoombox') the TERM signal 2015-03-31 15:58:39 MSG merovingian[1114]: database 'dw-zoombox' (14173) has exited with exit status 0 2015-03-31 15:58:39 MSG merovingian[1114]: database 'dw-zoombox' has shut down 2015-03-31 15:58:39 MSG control[1114]: (local): stopped database 'dw-zoombox' 2015-03-31 15:58:44 MSG control[1114]: (local): served status list 2015-03-31 15:58:44 MSG merovingian[1114]: starting database 'dw-zoombox', up min/avg/max: 2d/9w/18w, crash average: 0.00 0.10 0.03 (3-2=1) 2015-03-31 15:58:45 MSG dw-zoombox[19366]: arguments: /usr/bin/mserver5 --dbpath=/dados/dbfarm/dw-zoombox --set merovingian_uri=mapi:monetdb://SRVBIL01:50000/dw-zoombox --set mapi_open=false --set mapi_port=0 --set mapi_usock=/dados/dbfarm/dw-zoombox/.mapi.sock --set monet_vault_key=/dados/dbfarm/dw-zoombox/.vaultkey --set gdk_nr_threads=5 --set max_clients=64 --set sql_optimizer=default_pipe --set monet_daemon=yes 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # MonetDB 5 server v11.15.17 "Feb2013-SP5" 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Serving database 'dw-zoombox', using 5 threads 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Compiled for x86_64-pc-linux-gnu/64bit with 64bit OIDs dynamically linked 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Found 11.765 GiB available main-memory. 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Copyright (c) 1993-July 2008 CWI. 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Copyright (c) August 2008-2013 MonetDB B.V., all rights reserved 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Visit http://www.monetdb.org/ for further information 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # Listening for UNIX domain connection requests on mapi:monetdb:///dados/dbfarm/dw-zoombox/.mapi.sock 2015-03-31 15:59:34 MSG dw-zoombox[19366]: # MonetDB/SQL module loaded 2015-03-31 15:59:34 MSG control[1114]: (local): started 'dw-zoombox' 2015-03-31 15:59:34 MSG merovingian[1114]: proxying client 192.168.1.7:49105 for database 'dw-zoombox' to mapi:monetdb:///dados/dbfarm/dw-zoombox/.mapi.sock?database=dw-zoombox 2015-03-31 15:59:34 MSG merovingian[1114]: target connection is on local UNIX domain socket, passing on filedescriptor instead of proxying 2015-03-31 16:05:00 MSG merovingian[1114]: proxying client 192.168.1.7:49117 for database 'dw-zoombox' to mapi:monetdb:///dados/dbfarm/dw-zoombox/.mapi.sock?database=dw-zoombox 2015-03-31 16:05:00 MSG merovingian[1114]: target connection is on local UNIX domain socket, passing on filedescriptor instead of proxying 2015-03-31 16:05:20 MSG merovingian[1114]: proxying client 192.168.1.7:49119 for database 'dw-zoombox' to mapi:monetdb:///dados/dbfarm/dw-zoombox/.mapi.sock?database=dw-zoombox 2015-03-31 16:05:20 MSG merovingian[1114]: target connection is on local UNIX domain socket, passing on filedescriptor instead of proxying 2015-03-31 16:06:05 ERR discovery[1114]: cannot retrieve name info: Temporary failure in name resolution 2015-03-31 16:16:05 ERR discovery[1114]: cannot retrieve name info: Temporary failure in name resolution 2015-03-31 16:16:16 MSG merovingian[1114]: proxying client 192.168.1.8:50220 for database 'dw-zoombox' to mapi:monetdb:///dados/dbfarm/dw-zoombox/.mapi.sock?database=dw-zoombox 2015-03-31 16:16:16 MSG merovingian[1114]: target connection is on local UNIX domain socket, passing on filedescriptor instead of proxying 2015-03-31 16:17:04 MSG merovingian[1114]: proxying client 192.168.1.7:49159 for database 'dw-zoombox' to mapi:monetdb:///dados/dbfarm/dw-zoombox/.mapi.sock?database=dw-zoombox 2015-03-31 16:17:04 MSG merovingian[1114]: target connection is on local UNIX domain socket, passing on filedescriptor instead of proxying > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 2 Apr 2015 23:16:39 +0200 > From: Ying Zhang <Y.Zhang(a)cwi.nl> > To: "Communication channel for developers of the MonetDB suite." > <developers-list(a)monetdb.org> > Cc: users-list(a)monetdb.org > Subject: Re: MonetDB, data loss in restart server > Message-ID: <6C2903C0-4399-462D-9D3F-822351DE1979(a)cwi.nl> > Content-Type: text/plain; charset=windows-1252 > > Hello Luciano, > > Thanks for using monetdb, and sorry for the inconvenience. > Would you please give us some background information about your setting: > > - which OS? > - hardware spec. of your system > - which monetdb version? > - what does your query do? > - when and how did the database stop? If the monetdb server stopped, did it give any error message? > - what's at the end of the merovingian log? > > With kind regards, > > Jennie > >> On Apr 01, 2015, at 06:46, Luciano Sasso <luciano(a)gsgroup.com.br> wrote: >> >> Hello, >> >> I have used the MonetDB in production for some time, and I have nothing to complain about, >> are tables with more than 30 columns and 30 million records, do analysis through cubes in Pentaho / Saiku. >> Have restarted the server several times, and never had problems with loss of data.But today, >> after performing a query that was consuming a lot of machine, stopped the database, and when I started, >> data from some columns were blank. How can it, will the data were still uncommitted? >> when performing a query the MonetDB changes the data or reorganized? which may have occurred, I am concerned. >> >> >> -- >> <Slice_01.png> <Slice_02.jpg> >> Luciano Sasso Vieira >> Gestor - Intelig?ncia de Neg?cios >> >> luciano(a)gsgroup.com.br | tel: 17 3353-0833 | cel: 17 9706-9335 >> <Slice_04.png> www.gsgroup.com.br >> <Slice_06.png> >> >> >> Este email foi escaneado pelo Avast antiv?rus. >> www.avast.com >> >> >> _______________________________________________ >> developers-list mailing list >> developers-list(a)monetdb.org >> https://www.monetdb.org/mailman/listinfo/developers-list > > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > users-list mailing list > users-list(a)monetdb.org > https://www.monetdb.org/mailman/listinfo/users-list > > > ------------------------------ > > End of users-list Digest, Vol 32, Issue 6 > ***************************************** > > -- Luciano Sasso Vieira Gestor - Inteligência de Negócios luciano(a)gsgroup.com.br <http://www.gsgroup.com.br> | tel: 17 3353-0833 | cel: 17 9706-9335 www.gsgroup.com.br <http://www.gsgroup.com.br> --- Este email foi escaneado pelo Avast antivírus. http://www.avast.com

2 1

mserver5 segfault at 0 ip error 4 in lib_sql.so
by Caroptions Caroptions 08 Apr '15

08 Apr '15

Hi User-list! I am experiencing segfault monetdb crash, platform: CentOS Linux release 7.1.1503 (Core) 3.10.0-229.1.2.el7.x86_64 #1 SMP Fri Mar 27 03:04:26 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux Table: CREATE TABLE "marketdata"."quotes" ( "id" INTEGER NOT NULL DEFAULT next value for "marketdata"."seq_6391", "qtime" TIMESTAMP WITH TIME ZONE NOT NULL, "sdate" TIMESTAMP WITH TIME ZONE, "sym" VARCHAR(10) NOT NULL, "cur" VARCHAR(10) NOT NULL, "open" DOUBLE NOT NULL, "high" DOUBLE NOT NULL, "low" DOUBLE NOT NULL, "close" DOUBLE NOT NULL, "volume" DOUBLE, CONSTRAINT "quotes_id_pkey" PRIMARY KEY ("id") ); SQL: WITH cte(qtime, open, close, sdate, id, rnk) as ( select qtime, sdate, id, row_number() over (partition by sdate order by qtime asc) as rnk from marketdata.quotes where sym='SPY') select * from cte where rnk=1; ~2M records loaded from csv. Always segfault crashes monetdb daemon and DB became not repairable. Reproducible on MonetDB-11.19.9 installed from yum repository, as well as compiled from sources, with-optimization and without. Also it crashed on latest development source branch MonetDB-11c0c6cf614f with optimization and without. >From mclient and JDBC client. Any help is very appreciated!! I am available to debug / provide additional information if needed. Thanks, John

2 2

RE: MonetDB, data loss in restart server
by Luciano Sasso 08 Apr '15

08 Apr '15

OK, to update, just use apt-get, I will have problems with structure of database? I can use the old dbfarm, will be compatible? On Wed, Apr 08, 2015 at 11:08:53AM -0300, Luciano Sasso wrote: >/ Anyone have any idea about ? />/ />/ />/ - OS is Debian 7 64-bit, in a VM in VMware virtualization System. />/ - Hard disk 100G with 26% usage. />/ - 5 cores. />/ - 6G RAM. />/ - Version MonetDB: MonetDB 5 server v11.15.17 "Feb2013-SP5" /Thats a very old version. I recall a fix which we did in the meantime related to possible data loss, ie we added extra f(data)sync calls. Which could very well be the problem on a virtual system. Niels >/ - Table: create table (id int, dt date, product varchar (50) .... etc, />/ and another 30 columns between int and varchar. />/ - Query: select dt, count (id) from table where produtct in (....) group />/ by dt />/ / -- Luciano Sasso Vieira Gestor - Inteligência de Negócios luciano(a)gsgroup.com.br <http://www.gsgroup.com.br> | tel: 17 3353-0833 | cel: 17 9706-9335 www.gsgroup.com.br <http://www.gsgroup.com.br> --- Este email foi escaneado pelo Avast antivírus. http://www.avast.com

2 1

Incorrect results on joining with same table
by Vijay Krishna 08 Apr '15

08 Apr '15

Hi, I am using MonetDB Jan2014 SP1, running on MacOS. I have a table like this. sql>select * from two; +------+---------------------------------------------------------------------------------------------- | id | name | istrue | joindate | jointime | count | +======+==================================================== | 1 | null | false | 2017-12-08 | 2014-06-30 14:05:31.000000 | 2 | | 2 | vijay | false | 2017-12-08 | 2014-06-30 14:05:59.000000 | 3 | | 3 | krish | true | 2017-12-08 | 2014-06-30 14:06:17.000000 | 5 | | 4 | bat | true | null | null | null | | 5 | gotham | false | null | null | null | | 6 | wayne | false | null | null | 3 | +------+------------------------------------------------------------------------------------------------- 6 tuples (3.504ms) I tried to do a join on this table with the same table with the following query and got this result. sql>select two.id, two.name, two.count, twoAlias.id as id2, twoAlias.name as name2 from two left join two twoAlias on two.count=twoAlias.id; +------+------------------------------------------- -------------------------------------------------------+ | id | name | count | id2 | name2 +=============================================================+ | 1 | null | 2 | 2 | vijay | | 2 | vijay | 3 | 3 | krish | | 3 | krish | 5 | 5 | gotham | | 4 | bat | null | 3 | krish | | 5 | gotham | null | null | null | | 6 | wayne | 3 | null | null | +------+-------------------------------- --------------------------------------------------------------------+ 6 tuples (5.605ms) As it could be seen clearly, *this join is clearly resulting in a wrong output* at the 4th row. *The join criterion two.count=twoAlias.id is satisfied in the first 3 rows, but are clearly violated in the 4th row*, where in the values in the join columns (count and id2 here) are clearly found to be different. Hence the join rows are not matching. I expected that the output would be +------+------------------------------------------- -------------------------------------------------------+ | id | name | count | id2 | name2 +=============================================================+ | 1 | null | 2 | 2 | vijay | | 2 | vijay | 3 | 3 | krish | | 3 | krish | 5 | 5 | gotham | | 4 | bat | null | null | null | | 5 | gotham | null | null | null | | 6 | wayne | 3 | null | null | +------+-------------------------------- --------------------------------------------------------------------+ 6 tuples (5.605ms) Is this really a bug?? Else, what would be the logic behind this? Thanks in advance. With Regards, Vijayakrishna.P. Mobile : (+91) 9500402305.

2 2

RE: MonetDB, data loss in restart server
by Luciano Sasso 08 Apr '15

08 Apr '15

Anyone have any idea about ? - OS is Debian 7 64-bit, in a VM in VMware virtualization System. - Hard disk 100G with 26% usage. - 5 cores. - 6G RAM. - Version MonetDB: MonetDB 5 server v11.15.17 "Feb2013-SP5" - Table: create table (id int, dt date, product varchar (50) .... etc, and another 30 columns between int and varchar. - Query: select dt, count (id) from table where produtct in (....) group by dt Message: 1 Date: Thu, 2 Apr 2015 23:16:39 +0200 From: Ying Zhang<Y.Zhang at cwi.nl <https://www.monetdb.org/mailman/listinfo/users-list>> To: "Communication channel for developers of the MonetDB suite." <developers-list at monetdb.org <https://www.monetdb.org/mailman/listinfo/users-list>> Cc:users-list at monetdb.org <https://www.monetdb.org/mailman/listinfo/users-list> Subject: Re: MonetDB, data loss in restart server Message-ID:<6C2903C0-4399-462D-9D3F-822351DE1979 at cwi.nl <https://www.monetdb.org/mailman/listinfo/users-list>> Content-Type: text/plain; charset=windows-1252 Hello Luciano, Thanks for using monetdb, and sorry for the inconvenience. Would you please give us some background information about your setting: - which OS? - hardware spec. of your system - which monetdb version? - what does your query do? - when and how did the database stop? If the monetdb server stopped, did it give any error message? - what's at the end of the merovingian log? With kind regards, Jennie -- Luciano Sasso Vieira Gestor - Inteligência de Negócios luciano(a)gsgroup.com.br <http://www.gsgroup.com.br> | tel: 17 3353-0833 | cel: 17 9706-9335 www.gsgroup.com.br <http://www.gsgroup.com.br> --- Este email foi escaneado pelo Avast antivírus. http://www.avast.com

2 1

MonetDB for DW
by Luciano Sasso 08 Apr '15

08 Apr '15

Which of you with experience MonetDB production with large volume of data? You use some NoSQL base for storage of DW, and makes loads for MonetDB, for analysis, or use the monetdDB with dw too? Att, -- Luciano Sasso Vieira Gestor - Inteligência de Negócios luciano(a)gsgroup.com.br <http://www.gsgroup.com.br> | tel: 17 3353-0833 | cel: 17 9706-9335 www.gsgroup.com.br <http://www.gsgroup.com.br> --- Este email foi escaneado pelo Avast antivírus. http://www.avast.com

1 0

Cluster, How work ?
by Luciano Sasso 08 Apr '15

08 Apr '15

On a cluster using MonetDB, the documentation is not clear how it works. Data is automatically replicated between cluster nodes? There is an automatic load balancing? There sharding? Att, -- Luciano Sasso Vieira Gestor - Inteligência de Negócios luciano(a)gsgroup.com.br <http://www.gsgroup.com.br> | tel: 17 3353-0833 | cel: 17 9706-9335 www.gsgroup.com.br <http://www.gsgroup.com.br> --- Este email foi escaneado pelo Avast antivírus. http://www.avast.com

1 0

Sorting cluster's level queries
by José Raúl Pérez Rodríguez 06 Apr '15

06 Apr '15

Hi, I am new in monetdb. I was quering a multiplex funnel database and obviously (after read the description of multiplex funnel), the return values don't show up sorted. My question is: if using multiplex funnel databases queries can't be sorted, then how are sorted the queries at cluster level, i.e queries that retrieve data from 2 or more nodes? and how to perform that kind of quieries (not using a multiplex funnel). Thanks,

1 0

About "Zero-copy integration between MonetDB and R"
by Aris Setyawan 06 Apr '15

06 Apr '15

Hi all, First, thanks for implementing multi-core, parallel execution engine, ini MonetDB. This make MonetDB, the only opensource dbms, which have this feature. I'm sure, since MonetDB use mmap file, it is already use Zero-copy memory technique when loading data into memory for further processing. But, why the "Zero-copy integration between MonetDB and R" is not implemented in MonetDB release? -Aris

2 1

Building MonetDB from RPM on ROSA Desktop R4
by Brian Hood 05 Apr '15

05 Apr '15

Hi Guys, I know this isn't a support distro at all i'm just trying to get MonetDB to work on my ROSA Desktop R4 machine which is Linux / KDE with a Twist distro. http://pastebin.com/wiH6RPzQ Its all about this at the bottom. 1. cd . && /bin/sh ./config.status rpm.mk 2. config.status: creating rpm.mk 3. cd . && /bin/sh /home/brian/Downloads/rpmbuild/BUILD/MonetDB-11.19.7/buildtools/conf/missing automake-1.13 --foreign 4. configure.ac:372: error: required file 'buildtools/conf/compile' not found 5. configure.ac:372: 'automake --add-missing' can install 'compile' 6. Makefile:521: recipe for target 'Makefile.in' failed 7. make: *** [Makefile.in] Error 1 8. error: Bad exit status from /var/tmp/rpm-tmp.25756 (%build) 9. Automake the wrong version i'm not sure it seems to be upto date. Anybody any ideas on the off chance somebody has seen this before ? Regards, Brian Hood

2 2