- 07 Jan, 2014 1 commit
-
-
Sebastien Ponce authored
-
- 17 Dec, 2013 1 commit
-
-
Giuseppe Lo Presti authored
-
- 15 Nov, 2013 1 commit
-
-
Giuseppe Lo Presti authored
Further fixes on top of previous commit, still related to SR #140319. We can't ensure no one creates entities connected to a CastorFile as in particular draining and rebalancing activities may create many Disk2DiskCopyJob entries in bulk without taking locks on the corresponding files. Thus we have to compensate to constraint violation errors.
-
- 22 Oct, 2013 1 commit
-
-
Giuseppe Lo Presti authored
-
- 19 Sep, 2013 1 commit
-
-
Giuseppe Lo Presti authored
-
- 17 Sep, 2013 1 commit
-
-
Giuseppe Lo Presti authored
is to be used in all queries. This is to prevent Oracle from using a FAST FULL SCAN, which in fact is less fast as observed during stress tests.
-
- 08 Aug, 2013 1 commit
-
-
Sebastien Ponce authored
-
- 06 Aug, 2013 1 commit
-
-
Sebastien Ponce authored
-
- 01 Aug, 2013 1 commit
-
-
Sebastien Ponce authored
Improved draining by implementing back pressure on the scheduling (no more than 1000 jobs pending per drainingJob) and ordering of files by importance (no tape and fewer copies on disk first)
-
- 26 Jul, 2013 1 commit
-
-
Giuseppe Lo Presti authored
-
- 12 Jun, 2013 1 commit
-
-
Sebastien Ponce authored
This includes : - merge of d2dtransfer into diskmanagerd - drop of the WAITDISKTODISKCOPY state in DiskCopy, and of the diskcopies for ongoing replication in general. Now DiskCopies are only created at the end of the replication, when they actually exist on disk - drop of the StageReplicaRequest table and concept - introduction of the DiskToDiskCopyJob table, holding the list of ongoing disk to disk copies - a split of transferToSchedule into userTransferToSchedule and D2dTransferToSchedule - an according split in the dispatcher part of the transfer manager to call both methods. Note that with this commit, both methods fill the same FIFO queue of transfers and both take everything they can find without any throttling. At this stage, the draining facility is broken.
-
- 17 May, 2013 1 commit
-
-
Sebastien Ponce authored
Merged STAGED and CANBEMIGR into VALID in DiskCopy statuses and create a tapeStatus on the CastorFile with possible values NOTONTAPE, ONTAPE and DISKONLY. Note that user interface is not touched and still displays CANBEMIGR and STAGED. Only internals (particularly the DB) are concerned.
-
- 28 Mar, 2013 1 commit
-
-
Giuseppe Lo Presti authored
-
- 25 Feb, 2013 1 commit
-
-
Giuseppe Lo Presti authored
Implemented a new tool, deletediskcopy, to replace cleanLostFiles and provide all features described in the above bug and its related ones.
-
- 30 Jan, 2013 1 commit
-
-
Giuseppe Lo Presti authored
- Introduced support for a READONLY state for Diskservers and Filesystems. Hardware in such a state will accept migrations, [PrepareTo]Get and DiskCopyReplica requests, and will not accept recalls or [PrepareTo]Put/Update ones when they switch to write mode. - Dropped the {DiskServer,FileSystem}.adminStatus fields and the obsoleted code around them and introduced a new DiskServer.hwOnline boolean field, which is updated by the heartbeat reports and displayed by printdiskserver. In turn, modifydiskserver can not change the online flag but only the status. For backward compatibility, on stager_qry -s a Diskserver will be reported as DISABLED if hwOnline = 0 regardless its status.
-
- 25 Oct, 2012 1 commit
-
-
Giuseppe Lo Presti authored
-
- 24 Oct, 2012 1 commit
-
-
Sebastien Ponce authored
-
- 18 Oct, 2012 2 commits
-
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
- 17 Oct, 2012 1 commit
-
-
Giuseppe Lo Presti authored
Fixed bug #98250: Incorrect loop in archiveOrFailRepackSubreq leading to PL/SQL: numeric or value error. Fixed also a few more occurrences of FIRST..LAST, even though it appears the logic around wouldn't have led to the bug.
-
- 05 Oct, 2012 1 commit
-
-
Giuseppe Lo Presti authored
Added new GCTYPE_OVERWRITTEN constant and used it in recreateCastorFile to catch when files get overwritten. Side refactoring: all GCTYPE constants were renamed and added to the dconst package, as well as all draining processes specific constants.
-
- 27 Sep, 2012 2 commits
-
-
Giuseppe Lo Presti authored
-
Giuseppe Lo Presti authored
-
- 18 Sep, 2012 1 commit
-
-
Giuseppe Lo Presti authored
Fixed bug #96943: Inserts without columns and those using select * sources need to be replaced by statements that explicitly name columns
-
- 25 Jul, 2012 1 commit
-
-
Sebastien Ponce authored
Made GcPolicy mandatory in the SvcClass table (not null + foreign key on GcPolicy name) + recreated SQL scripts
-
- 25 Jun, 2012 1 commit
-
-
Giuseppe Lo Presti authored
The creation of the index took 240 secs with 110M files.
-
- 25 May, 2012 1 commit
-
-
Giuseppe Lo Presti authored
Introduced a generic procedure to run DB jobs and log any exception they would throw with its full stack trace.
-
- 29 Mar, 2012 1 commit
-
-
Sebastien Ponce authored
Used it to put back the migration related messages in the tapegatewayd log
-
- 22 Mar, 2012 1 commit
-
-
Sebastien Ponce authored
Fixed timestamp issues with the dedicated cleanup logging facility. Extended the DLF api with a method taking a message string rather than a message number. This should be the only used from now on.
-
- 21 Mar, 2012 1 commit
-
-
Giuseppe Lo Presti authored
-
- 26 Jan, 2012 1 commit
-
-
Giuseppe Lo Presti authored
-
- 30 Nov, 2011 1 commit
-
-
Giuseppe Lo Presti authored
That index is nowadays useless with >99% of DiskCopies in status 0.
-
- 29 Nov, 2011 1 commit
-
-
Giuseppe Lo Presti authored
-
- 24 Nov, 2011 1 commit
-
-
Giuseppe Lo Presti authored
to drop Repack requests when they are just inserted without SubRequests (Only case in the request handling logic). The solution is not perfect though, there's a race between the cleaning job being too aggressive and the NS answering too slowly for a Repack request to start getting populated before the timeout. To be improved.
-
- 22 Sep, 2011 1 commit
-
-
Sebastien Ponce authored
The main gains are : - simpler DB schema : StreamToTapeCopy and SvcClass2TapePool have disappeared. TapeCopy is split in to the dedicated RecallJob and MigrationJob. Stream is replaced by MigrationMount that includes all data needed from the Tape table (so the Tape table only used by recall now). - simpler concepts : MigrationMounts match with mounts and do not survive them (as Stream did), MigrationRouting table explicitely gives the migration routing. NbDrives has moved to the TapePool - less components : migHunter is gone, replaced by the MigrationRouting table and by a DB job that starts the needed mounts - better error reporting : migration failures due to bad configuration are reported sunchronously to the user, on file close or even on file opening when possible. e.g. if the combination of svcClass and fileClass given by the the user is not present in the MigrationRouting table, the file opening will fail with message "File recreation canceled since the file cannot be routed to tape". This case used to loop forever in the migration part. Conflicts: castor/db/oracleCommon.sql castor/db/oracleSchema.sql castor/db/oracleTapeGateway.sql castor/tape/tapegateway/daemon/TapeStreamLinkerThread.cpp castor/tape/tapegateway/daemon/VmgrTapeGatewayHelper.cpp castor/tape/tapegateway/daemon/VmgrTapeGatewayHelper.hpp castor/tape/tapegateway/daemon/ora/OraTapeGatewaySvc.cpp castor/tape/tapegateway/daemon/ora/OraTapeGatewaySvc.hpp codeGeneration/RH.xmi Conflicts: castor/db/oracleCommon.sql
-
- 29 Jun, 2011 1 commit
-
-
Sebastien Ponce authored
Improved GC efficiency by creating a new index on Diskcopy based primarily on filesystem and gcweight and by throttling the number of (heavy) GC requests running concurrently
-
- 10 Jun, 2011 1 commit
-
-
Giuseppe Lo Presti authored
the logic around it from the CASTOR schemata) for the stager db. This involved the usage of SubRequest.reqType where needed, and a major change of subRequestToDo and subRequestFailedToDo, which now return the corresponding Request and Client data in one go for the stager processing. Furthermore, fillObj() calls where the remote entity is an abstract class (like OBJ_Request) have been dropped, and (one) createObj() call without type has been amended.
-
- 06 Jun, 2011 3 commits
-
-
Dennis Waldron authored
-
Dennis Waldron authored
-
Dennis Waldron authored
-