- Jun 12, 2012
-
-
Sebastien Ponce authored
Better implementation of the CastorConfiguration class. No inheritance from map anymore, simpler locks, and proper checks that the lock taking succeed
-
Sebastien Ponce authored
-
- Jun 11, 2012
-
-
Giuseppe Lo Presti authored
-
Giuseppe Lo Presti authored
-
Giuseppe Lo Presti authored
scripts for both the Stager and the NS databases, in order to keep the other scripts idempotent. Note that a number of bootstrap insertions are still missing from the Stager db (e.g. tapepools, routes), will eventually come later.
-
Sebastien Ponce authored
-
- Jun 08, 2012
-
-
Giuseppe Lo Presti authored
-
Giuseppe Lo Presti authored
-
Andrea Ieri authored
Added chmod_invalidNumericTooLong to mirror mkdir_invalidModeTooLong Fixed three more tests which had a useless check for $?
-
Andrea Ieri authored
ns/Cns_procreq.c: Added checks on unmarshalled mode to make sure it isn't an invalid number (no more blind trust on client). ns/Cns_mkdir.c and ns/Cns_chmod.c: Removal of silent truncating of specified mode via bitmask and substitution with explicit checking. The library aborts if an invalid mode is found instead of "fixing" the input. ns/nsmkdir.c: Added explicit validation of submitted mode. It wouldn't be strictly needed as the server wouldn't allow invalid modes anyway now, but it allows for the printing of a specific error message and avoids useless communications with the nameserver.
-
- Jun 06, 2012
-
-
Steven Murray authored
bug #92460: tapebridged should gracefully shutdown a migration tape-session when tapegatewayd reports a disabled tape Fixed bad behaviour of tapebridged when it has to cope with a DISABLED tape just before its request for the first file to migrate.
-
Steven Murray authored
bug #92460: tapebridged should gracefully shutdown a migration tape-session when tapegatewayd reports a disabled tape Tests for the 2.1.13-0 release have shown that the tapebridged daemon does not cope correctly with a tape being DISABLED just before it asks for the first file to be migrated. The tapebridged daemon incorreectly ends the tape session with an EndNotification message when it should end the tape session with a EndNotificationErrorReport. This commit adds the unit-test that fails if it detects this bad behaviour.
-
Eric Cano authored
Added support for wiping and re-creating the name server in a dev environement. Followed up with the development of the 2.1.13 version.
-
Eric Cano authored
Ported missing synonym creation from upgrade script to creation script for name server. Added the creation of the root entry in the same place. Ported missing creation of the default recallGroup in the stager DB (in tape gateway section).
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
- Jun 05, 2012
-
-
Giuseppe Lo Presti authored
-
Sebastien Ponce authored
-
Sebastien Ponce authored
Fixed a race condition between renewal of configuration and the check whether we should renew that looked at the config (for the timeout) outside of a read lock
-
Eric Cano authored
Removed unreferenced file.
-
Sebastien Ponce authored
Fixed stupid mistake by which all tapes were requested to VDQM in read mode. The scarry thing here is that it did not prevent migration to happen !
-
- Jun 04, 2012
-
-
Giuseppe Lo Presti authored
-
Giuseppe Lo Presti authored
-
Eric Cano authored
-
- Jun 01, 2012
-
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
- May 31, 2012
-
-
Sebastien Ponce authored
-
Sebastien Ponce authored
Added indexes and comment to prevent table locks created by foreign key constraints and explain the situation
-
Sebastien Ponce authored
-
- May 30, 2012
-
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-
Giuseppe Lo Presti authored
-
Giuseppe Lo Presti authored
used in conjunction with distributed transactions. Updated upgrade scripts.
-
Giuseppe Lo Presti authored
endTapeSession be called in that context, and that the tapebridged should handle the case of receiving a 'no such session/mount' error if it asks for more work on a session that had to be prematurely failed.
-
Sebastien Ponce authored
-
Sebastien Ponce authored
-