230 likes | 336 Vues
ATLAS DDM Operations - III DPM at T2’s. Jiří Chudoba ATLAS meeting , 25 . 9 .2007, CNAF. DPM at T2’s. FZK cloud: FZU srm://golias100.farm.particle.cz/dpm/farm.particle.cz/home/atlas/ CYF srm://dpm.cyf-kr.edu.pl/dpm/cyf-kr.edu.pl/home/atlas/ SARA cloud:
E N D
ATLAS DDM Operations - IIIDPM at T2’s Jiří Chudoba ATLAS meeting, 25.9.2007, CNAF
DPM at T2’s • FZK cloud: • FZU srm://golias100.farm.particle.cz/dpm/farm.particle.cz/home/atlas/ • CYF srm://dpm.cyf-kr.edu.pl/dpm/cyf-kr.edu.pl/home/atlas/ • SARA cloud: • NIKHEF: srm://tbn18.nikhef.nl/dpm/nikhef.nl/home/atlas/ • IHEP: srm://dpm0001.m45.ihep.su/dpm/m45.ihep.su/home/atlas/ • PNPI: srm://cluster.pnpi.nw.ru/dpm/pnpi.nw.ru/home/atlas/dq2 • CNAF cloud: • Milano: srm://grid015.mi.infn.it/dpm/mi.infn.it/home/atlas/ • Roma1: srm://grid-cert-03.roma1.infn.it/dpm/roma1.infn.it/home/atlas/ • Napoli: srm://t2-dpm-01.na.infn.it/dpm/na.infn.it/home/atlas/ • LNF: srm://atlasse.lnf.infn.it/dpm/lnf.infn.it/home/atlas/ ATLAS DDM Operations
DPM • DPM documentation: • https://twiki.cern.ch/twiki/bin/view/LCG/DataManagementDocumentation • Description of DBs on some wiki pages: http://www.gridpp.ac.uk/wiki/DPM_MySQL_database • DBs: cns_db, dpm_db • Tables from cns_db: Cns_file_metadata, Cns_file_replica, Cns_groupinfo, Cns_userinfo ATLAS DDM Operations
Cns_file_metadata +---------------+---------------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +---------------+---------------------+------+-----+---------+----------------+ | rowid | bigint(20) unsigned | | PRI | NULL | auto_increment | | fileid | bigint(20) unsigned | YES | MUL | NULL | | | parent_fileid | bigint(20) unsigned | YES | MUL | NULL | | | guid | varchar(36) | YES | MUL | NULL | | | name | varchar(255) | YES | | NULL | | | filemode | int(10) unsigned | YES | | NULL | | | nlink | int(11) | YES | | NULL | | | owner_uid | int(10) unsigned | YES | | NULL | | | gid | int(10) unsigned | YES | | NULL | | | filesize | bigint(20) unsigned | YES | | NULL | | | atime | int(11) | YES | | NULL | | | mtime | int(11) | YES | | NULL | | | ctime | int(11) | YES | | NULL | | | fileclass | smallint(6) | YES | | NULL | | | status | char(1) | YES | | NULL | | | csumtype | char(2) | YES | | NULL | | | csumvalue | varchar(32) | YES | | NULL | | | acl | blob | YES | | NULL | | +---------------+---------------------+------+-----+---------+----------------+ ATLAS DDM Operations
Cns_file_replica +------------+---------------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +------------+---------------------+------+-----+---------+----------------+ | rowid | bigint(20) unsigned | | PRI | NULL | auto_increment | | fileid | bigint(20) unsigned | YES | MUL | NULL | | | nbaccesses | bigint(20) unsigned | YES | | NULL | | | atime | int(11) | YES | | NULL | | | ptime | int(11) | YES | | NULL | | | status | char(1) | YES | | NULL | | | f_type | char(1) | YES | | NULL | | | poolname | varchar(15) | YES | | NULL | | | host | varchar(63) | YES | MUL | NULL | | | fs | varchar(79) | YES | | NULL | | | sfn | blob | YES | MUL | NULL | | | ctime | int(11) | YES | | NULL | | | ltime | int(11) | YES | | NULL | | | r_type | char(1) | YES | | NULL | | | setname | varchar(36) | YES | | NULL | | +------------+---------------------+------+-----+---------+----------------+ ATLAS DDM Operations
Cns_userinfo +----------+------------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +----------+------------------+------+-----+---------+----------------+ | rowid | int(10) unsigned | | PRI | NULL | auto_increment | | userid | int(11) | YES | | NULL | | | username | varchar(255) | YES | MUL | NULL | | +----------+------------------+------+-----+---------+----------------+ ATLAS DDM Operations
Cns_groupinfo +-----------+------------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +-----------+------------------+------+-----+---------+----------------+ | rowid | int(10) unsigned | | PRI | NULL | auto_increment | | gid | int(11) | YES | | NULL | | | groupname | varchar(255) | YES | MUL | NULL | | +-----------+------------------+------+-----+---------+----------------+ ATLAS DDM Operations
ATLAS groups Examples from DPM at golias100 select * from Cns_groupinfo where groupname like '%atlas%'; +-------+------+-------------------------------+ | rowid | gid | groupname | +-------+------+-------------------------------+ | 1 | 1309 | atlassgm | | 3 | 1308 | atlas | | 7 | 1307 | atlas/Role=lcgadmin | | 8 | 1311 | atlasprd | | 10 | 1310 | atlas/Role=production | | 16 | 116 | atlas/lcg1 | | 18 | 118 | atlas/usatlas | | 21 | 121 | atlas/ca | | 30 | 2695 | atlas/de | | 32 | 2697 | atlas/nl | | 33 | 2698 | atlas/fr | | 35 | 2700 | atlas/usatlas/Role=production | +-------+------+-------------------------------+ Until this summer, we mapped all ATLAS users to gid=1307 ATLAS DDM Operations
ATLAS users select owner_uid, username, sum(filesize)/1000000000 as size_GB from Cns_file_metadata, Cns_userinfo, Cns_groupinfo where owner_uid=userid and Cns_groupinfo.gid=C ns_file_metadata.gid and Cns_groupinfo.groupname like "atlas%" group by owner_uid; +-----------+-------------------------------------------------------------------------------------------------------------+---------+ | owner_uid | username | size_GB | +-----------+-------------------------------------------------------------------------------------------------------------+---------+ | 101 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Guido Negri/Email=guido.negri@mi.infn.it | 414.78 | | 102 | /DC=cz/DC=cesnet-ca/O=Institute of Physics of the Academy of Sciences of the CR/CN=Jiri Chudoba | 56.85 | | 104 | /C=CH/O=CERN/OU=GRID/CN=Simone Campana 7461 - ATLAS | 21.24 | | 106 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Silvia Resconi/Email=Silvia.Resconi@mi.infn.it | 95.37 | | 108 | /C=CH/O=CERN/OU=GRID/CN=Jiri Kosina 6222 | 0.00 | | 115 | /C=CA/O=Grid/OU=westgrid.ca/CN=Rodney Walker | 22.54 | | 128 | /C=UK/O=eScience/OU=Lancaster/L=Physics/CN=james catmore | 0.00 | | 132 | /O=GermanGrid/OU=LMU/CN=Gernot Krobath | 1.89 | | 146 | /O=GermanGrid/OU=LMU/CN=Cedric Serfon | 0.10 | | 149 | /O=GermanGrid/OU=LMU/CN=John Kennedy | 16.40 | | 154 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Guido Negri 2 | 139.78 | | 157 | /C=CH/O=CERN/OU=GRID/CN=Miguel De Oliveira Branco 2423 | 217.00 | | 160 | /C=CA/O=Grid/OU=phys.uvic.ca/CN=Ashok Agarwal | 1.12 | | 166 | /O=GermanGrid/OU=LMU/CN=Johannes Elmsheuser | 0.00 | | 189 | /C=PL/O=GRID/O=INP/CN=Andrzej Olszewski | 3.10 | | 191 | /C=TW/O=AS/OU=PHYS/CN=suijian zhou/Email=suijian.zhou@cern.ch | 2.29 | | 192 | /O=GermanGrid/OU=DESY/CN=Andreas Haupt | 1.11 | | 195 | /C=UK/O=eScience/OU=QueenMaryLondon/L=Physics/CN=kondo gnanvo | 0.00 | | 218 | /C=CH/O=CERN/OU=GRID/CN=Piotr Nyczyk 6217 - Atlas | 0.00 | | 219 | /C=CH/O=CERN/OU=GRID/CN=David Cameron 0841 | 453.53 | | 247 | /DC=ch/DC=cern/OU=Organic Units/OU=Users/CN=dcameron/CN=555105/CN=David Cameron | 0.00 | | 248 | /C=PL/O=GRID/O=IFJ/CN=Anna Kaczmarska - Atlas | 0.00 | | 250 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Davide Rebatto/Email=David.Rebatto@mi.infn.it | 44.54 | | 258 | /DC=ch/DC=cern/OU=Organic Units/OU=Users/CN=mlassnig/CN=663551/CN=Mario Lassnig | 4741.78 | | 263 | /DC=org/DC=doegrids/OU=People/CN=Sanjay Padhi 496075 | 155.73 | | 318 | /C=IT/O=INFN/OU=Personal Certificate/L=Roma 1/CN=Alessandro De Salvo/Email=Alessandro.DeSalvo@roma1.infn.it | 2.12 | | 334 | /C=UK/O=eScience/OU=Glasgow/L=Compserv/CN=caitriana nicholson | 2.14 | | 358 | /DC=ch/DC=cern/OU=Organic Units/OU=Users/CN=sch/CN=664224/CN=Ariel Gustavo Schwartzman | 0.03 | +-----------+-------------------------------------------------------------------------------------------------------------+---------+ ATLAS DDM Operations
ATLAS production users select owner_uid, username, Cns_groupinfo.groupname, sum(filesize)/1000000000 as size_GB from Cns_file_metadata, Cns_userinfo, Cns_groupinfo where owner_uid=useri d and Cns_groupinfo.gid=Cns_file_metadata.gid and Cns_groupinfo.gid=1310 group by owner_uid; +-----------+-------------------------------------------------------------------------------------------------+-----------------------+---------+ | owner_uid | username | groupname | size_GB | +-----------+-------------------------------------------------------------------------------------------------+-----------------------+---------+ | 106 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Silvia Resconi/Email=Silvia.Resconi@mi.infn.it | atlas/Role=production | 6.41 | | 115 | /C=CA/O=Grid/OU=westgrid.ca/CN=Rodney Walker | atlas/Role=production | 3.56 | | 154 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Guido Negri 2 | atlas/Role=production | 14.54 | | 250 | /C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Davide Rebatto/Email=David.Rebatto@mi.infn.it | atlas/Role=production | 14.85 | | 263 | /DC=org/DC=doegrids/OU=People/CN=Sanjay Padhi 496075 | atlas/Role=production | 71.11 | +-----------+-------------------------------------------------------------------------------------------------+-----------------------+---------+ select owner_uid, Cns_groupinfo.groupname, sum(filesize)/1000000000 as size_GB from Cns_file_metadata, Cns_userinfo, Cns_groupinfo where owner_uid=userid and Cns_groupinfo.gid=Cns_file_metadata.gid and Cns_groupinfo.gid=1310 group by owner_uid; +-----------+-----------------------+---------+ | owner_uid | groupname | size_GB | +-----------+-----------------------+---------+ | 106 | atlas/Role=production | 6.41 | | 115 | atlas/Role=production | 3.56 | | 154 | atlas/Role=production | 14.54 | | 250 | atlas/Role=production | 14.85 | | 263 | atlas/Role=production | 71.11 | +-----------+-----------------------+---------+ ATLAS DDM Operations
How old are ATLAS files select date_format(from_unixtime(ctime),'%Y-%m') as month, count(*) as nfiles, sum(filesize)/1000000000 as size_GB from Cns_file_metadata, Cns_groupinfo where Cns_groupinfo.gid=Cns_file_metadata.gid and Cns_groupinfo.groupname like "atlas%" group by month; +---------+--------+---------+ | month | nfiles | size_GB | +---------+--------+---------+ | 2006-03 | 1920 | 127.39 | | 2006-04 | 10718 | 859.46 | | 2006-05 | 23333 | 1645.98 | | 2006-06 | 5275 | 601.59 | | 2006-07 | 506 | 35.49 | | 2006-08 | 334 | 35.53 | | 2006-09 | 162 | 18.39 | | 2006-10 | 5414 | 665.23 | | 2006-11 | 39 | 0.00 | | 2006-12 | 973 | 0.00 | | 2007-01 | 51 | 0.00 | | 2007-02 | 159 | 0.00 | | 2007-03 | 1922 | 4.94 | | 2007-04 | 2336 | 72.66 | | 2007-05 | 7966 | 346.48 | | 2007-06 | 44197 | 1822.96 | | 2007-07 | 845 | 85.89 | | 2007-08 | 27705 | 1844.95 | | 2007-09 | 22093 | 1020.92 | +---------+--------+---------+ ATLAS DDM Operations
Access to files owned by production select nbaccesses, count(*) from Cns_file_replica, Cns_file_metadata where Cns_file_replica.fileid=Cns_file_metadata.fileid and Cns_file_metadata.gi d=1310 group by nbaccesses ; +------------+----------+ | nbaccesses | count(*) | +------------+----------+ | 1 | 978 | | 2 | 825 | | 3 | 183 | | 4 | 30 | | 5 | 30 | | 6 | 11 | | 7 | 5 | | 8 | 1 | | 10 | 2 | | 11 | 1 | | 12 | 2 | | 13 | 1 | | 14 | 2 | | 16 | 1 | | 19 | 2 | | 22 | 3 | | 30 | 1 | | 31 | 3 | | 32 | 5 | | 33 | 1 | | 67 | 1 | +------------+----------+ 21 rows in set (1.77 sec) 1 access = file was only stored – no copy on T1 ATLAS DDM Operations
Production files not yet replicated to T1 select md.name, date_format(from_unixtime(md.ctime),'%Y-%m-%d') as day from Cns_file_replica r, Cns_file_metadata md where r.fileid=md.fileid and md.gid=1310 and r.nbaccesses=1 order by day asc limit 10 ; +---------------------------------+------------+ | name | day | +---------------------------------+------------+ | log.012663._00127.job.log.tgz.1 | 2007-08-23 | | log.010685._80590.job.log.tgz.1 | 2007-08-23 | | ESD.011212._03771.pool.root.4 | 2007-08-29 | | EVNT.011675._00528.pool.root.2 | 2007-08-29 | | log.011680._00605.job.log.tgz.2 | 2007-08-29 | | log.013304._00016.job.log.tgz.3 | 2007-08-29 | | RDO.013306._00003.pool.root.1 | 2007-08-29 | | log.013303._00022.job.log.tgz.3 | 2007-08-29 | | RDO.013304._00061.pool.root.3 | 2007-08-29 | | HITS.013304._00014.pool.root.3 | 2007-08-29 | +---------------------------------+------------+ ATLAS DDM Operations
Usage of files replicated by DDM ATLAS DDM Operations
AOD Usage select count(*) from Cns_file_replica r where fileid in (select fileid from Cns_file_metadata where owner_uid=258 and name like '%AOD%'); 66436 AOD files owned by DDM (Mario’s proxy) select r.nbaccesses, count(*) from Cns_file_replica r where fileid in (select fileid from Cns_file_metadata where owner_uid=258 and name like '%AOD%') group by n baccesses limit 15; +------------+----------+ | nbaccesses | count(*) | +------------+----------+ | 1 | 45411 | | 2 | 11384 | | 3 | 3324 | | 4 | 1974 | | 5 | 1629 | | 6 | 755 | | 7 | 762 | | 8 | 402 | | 9 | 202 | | 10 | 188 | | 11 | 135 | | 12 | 70 | | 13 | 56 | | 14 | 34 | | 15 | 24 | +------------+----------+ 66 436 AOD files, 45411 never used via DPM, 90 files used more than 15 times ATLAS DDM Operations
dpm_db dpm_get_filereq: +-------------+---------------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +-------------+---------------------+------+-----+---------+----------------+ | rowid | bigint(20) unsigned | | PRI | NULL | auto_increment | | r_token | varchar(36) | YES | MUL | NULL | | | f_ordinal | int(11) | YES | | NULL | | | from_surl | blob | YES | MUL | NULL | | | lifetime | int(11) | YES | | NULL | | | f_type | char(1) | YES | | NULL | | | s_token | varchar(36) | YES | | NULL | | | flags | int(11) | YES | | NULL | | | server | varchar(63) | YES | | NULL | | | pfn | blob | YES | MUL | NULL | | | actual_size | bigint(20) unsigned | YES | | NULL | | | status | int(11) | YES | | NULL | | | errstring | varchar(255) | YES | | NULL | | | protocol | varchar(7) | YES | | NULL | | | ret_policy | char(1) | YES | | NULL | | +-------------+---------------------+------+-----+---------+----------------+ ATLAS DDM Operations
total transfers select sum(actual_size)/1000000000 as get_GB from dpm_get_filereq ; +----------+ | get_GB | +----------+ | 47163.37 | +----------+ select sum(actual_size)/1000000000 as put_GB from dpm_put_filereq ; +----------+ | put_GB | +----------+ | 22372.59 | +----------+ ATLAS DDM Operations
put put requests: select date_format(from_unixtime(ctime),'%Y-%m') as month, count(*) from dpm_req where r_token in (select r_token from dpm_put_filereq) group by month; +---------+----------+ | month | count(*) | +---------+----------+ | 2006-03 | 2549 | | 2006-04 | 16604 | | 2006-05 | 28188 | | 2006-06 | 9132 | | 2006-07 | 5342 | | 2006-08 | 5116 | | 2006-09 | 4370 | | 2006-10 | 34282 | | 2006-11 | 10414 | | 2006-12 | 15870 | | 2007-01 | 11820 | | 2007-02 | 10638 | | 2007-03 | 31518 | | 2007-04 | 21938 | | 2007-05 | 60971 | | 2007-06 | 70478 | | 2007-07 | 8267 | | 2007-08 | 61377 | | 2007-09 | 75111 | +---------+----------+ ATLAS DDM Operations
get Get requests: select date_format(from_unixtime(ctime),'%Y-%m') as month, count(*) from dpm_req where r_token in (select r_token from dpm_get_filereq) group by month; +---------+----------+ | month | count(*) | +---------+----------+ | 2006-03 | 1039 | | 2006-04 | 8063 | | 2006-05 | 34846 | | 2006-06 | 20797 | | 2006-07 | 9131 | | 2006-08 | 7142 | | 2006-09 | 18359 | | 2006-10 | 16221 | | 2006-11 | 13251 | | 2006-12 | 8350 | | 2007-01 | 3842 | | 2007-02 | 8873 | | 2007-03 | 13731 | | 2007-04 | 10503 | | 2007-05 | 7373 | | 2007-06 | 7551 | | 2007-07 | 9663 | | 2007-08 | 48762 | | 2007-09 | 41102 | +---------+----------+ ATLAS DDM Operations
DPM dumps • export DPNS_HOST=t2-dpm-01.na.infn.it • export DPM_HOST=t2-dpm-01.na.infn.it • dpns-ls -lR /dpm/na.infn.it/home/atlas > napoli.dpnsls.20070924.list ATLAS DDM Operations
CheckDpmConsistency • https://twiki.cern.ch/twiki/bin/view/LCG/CheckDpmConsistency • Author: Lana Abadie • The DPM consistency check tool is an admin tool which can be used for the following purposes • behaviour by default: • it allows listing files which are stored in the name server but not physically stored; • it allows listing files which are stored physically and not declared in the DPNS DB; • it allows listing files whose sizes stored in the name server differ from the physical size ATLAS DDM Operations
dpns-cleanup • By Lana Abadie • I tested it when we lost all files at 1 pool node • Uses these main calls: • dpns_listreplicax • dpns_delreplica • dpns_getpath • dpns_unlink • Example: ./dpns-cleanup --dpns_host_name golias100.farm.particle.cz --server se2.farm.particle.cz started at 11:06, finished 14:22 deleted 24414 files. ATLAS DDM Operations
ATLAS groups ATLAS DDM Operations