Commit Graph

235 Commits

Author SHA1 Message Date
Alex Huang 28963568cd Somehow I missed a problem with lock release that was fixed in other branches 2011-08-02 15:37:06 -07:00
frank 44f3ac98f6 A couple of things:
add 229 upgrade path

Bug 10860 - PremiumUpgrade is not run when upgrading to 2.2.8 (edit)

run checker before any components loaded
2011-07-29 20:05:50 -07:00
Alex Huang 1f4d34b9d5 print further into the stack so we can see who's calling the Transaction methods 2011-07-29 10:51:16 -07:00
frank 9aae53fbb3 Bug 10860 - PremiumUpgrade is not run when upgrading to 2.2.8
Use a new target "system-integrity-checker" in components.xml/components-premium.xml.
All checkers must be explicitly specified in XML file, they will execute before any components load

status 10860: resolved fixed
2011-07-27 17:43:38 -07:00
Sheng Yang b116f54741 Generate keystore using RSA rather than DSA
Also fix a typo.
2011-07-25 18:23:44 -07:00
Alex Huang b30c55a53f missing file 2011-07-25 18:07:17 -07:00
Alex Huang 9c627a15f3 Inaccurate clock new gets an mbean to control it 2011-07-25 16:01:31 -07:00
Alex Huang 10ac7753ed Switched ping to use the same db connection so that running out of db connections won't affect basic operations 2011-07-25 10:36:00 -07:00
Alex Huang ec4a46ece3 use in memory counter for locks so we don't do a delete everytime 2011-07-23 15:24:33 -07:00
Alex Huang 3f18192df8 Make all connections READ COMMITTED isolation level instead of setting it everytime we get the db connection causing useless round trips 2011-07-23 14:58:32 -07:00
Alex Huang 44ce9488a6 propagate lock table fixes 2011-07-22 11:30:23 -07:00
Sheng Yang 15bf729927 bug 10714: Implement packet fragmentation
Also add an simple nio unit test.

status 10714: resolved fixed
2011-07-22 10:08:40 -07:00
alena e344d5f63a bug 10759: fixed the bug in SqlGenerator - attributes weren't set right if GENERIC.DAO.REMOVED attribute was present in the list (happened for disk and service offerings)
status 10759: resolved fixed
2011-07-21 18:48:48 -07:00
Alex Huang f500a01511 more db connection controls through jmx 2011-07-21 13:38:53 -07:00
Alex Huang d7667180c5 bug 10816: more db lock controls 2011-07-21 13:38:37 -07:00
Alex Huang 5cfad0b6c9 connection conierge to deal with connections that are not returned to pools 2011-07-21 13:32:43 -07:00
Alex Huang 3ea6948645 bug 10716: Changed transaction level to read committed to allow for more concurrency 2011-07-20 15:54:51 -07:00
alena 25206e6257 bug 10681: for serviceOfferings, diskOfferings, templates/isos return records with not null Removed field on findById method call
status 10681: resolved fixed
2011-07-13 17:16:07 -07:00
Alex Huang b54edececf resolved a bug where the wrong msid was used in releasing locks 2011-07-13 09:52:35 -07:00
Alex Huang 5619f0490c added more locking logging 2011-07-13 09:29:38 -07:00
Sheng Yang a2de0f60d5 Move disconnect log to TRACE level 2011-07-11 11:08:40 -07:00
Sheng Yang 612f776892 Fix keystore loading path 2011-07-11 10:59:04 -07:00
Sheng Yang e8f317243f SSL: Fix leaking file descriptor
And bad connection fail handling.
2011-07-08 22:04:34 -07:00
Sheng Yang 0f09546d04 Fix Class cast issue 2011-07-08 20:34:12 -07:00
Sheng Yang 8ada2852bf Suppress connection error messages
Load balancers would trigger this error message again and again.. Put them in to
debug/trace level, instread of showing a calltrace.
2011-07-08 18:12:29 -07:00
Alex Huang 7e9836dfd0 Some operations on the lock table allowed through jmx 2011-07-06 16:09:05 -07:00
Alex Huang b45bc9449e bug 10548: This problem is actually still a running out of db connection problem. However, it points out a weakness in the Merovingian2 design. It should keep a database connection for itself. I originally intended to have that after writing the first edition but got distracted and didn't finish. Here in the new implementation, Merovingian2 allocates a database connection for itself and keeps it alive at 10 second intervals with a SELECT 1 to avoid the connection being considered as abandoned by the dbcp pool. 2011-07-03 19:16:24 -07:00
Alex Huang cfc25d01be bug 10501: This is really Kelven's bug but I'll fix it for him anyways. DAO code already have a way to extract the DB connection from a transaction that is stored in the TLS. There's no real reason for the DAO code to add special semantics to use a different DB connection. That can be done by simply switching the transaction before it even reached the dao code. Think about it. Why would anyone want to call one dao function, switch transaction, and then switch back. The right thing is for the caller to switch transaction, call a series of dao codes, and switch it back. That's the semantics I changed to. By doing this, it also eliminates the number of debug messages in this bug. 2011-07-01 11:03:12 -07:00
Alex Huang d5b3fe4e35 bug 10322: Transaction is using the new Merovingian 2011-06-29 15:15:52 -07:00
Alex Huang 81213c2be2 bug 10322: Transaction is using the new Merovingian 2011-06-29 12:35:30 -07:00
Alex Huang e16a013335 bug 10322: better mannered merovingians 2011-06-29 12:01:43 -07:00
Alex Huang 2219269e83 bug 10322: arrival of the next generation Merovingians 2011-06-29 12:01:43 -07:00
Alex Huang cbee5a6044 bug 10322: arrival of the next generation Merovingians 2011-06-29 12:01:43 -07:00
Kelven Yang 1edf772f92 bug 10480, 10494: NPE fix in VirtualMachineManagerImpl, move keystore upgrade sql to upgrade225to226.sql 2011-06-28 15:00:34 -07:00
Alex Huang 2f391d2b9d more jmx stuff 2011-06-28 01:17:10 -07:00
Alex Huang 90488eea7c more jmx stuff 2011-06-27 18:26:30 -07:00
Alex Huang b7e4298c94 Added JMX to Transaction 2011-06-27 17:24:26 -07:00
Alex Huang 97b562fc2a logging changes 2011-06-27 10:18:40 -07:00
Kelven Yang 3b93d353d6 bug 9605: use CIDR from default interface 2011-06-23 17:48:37 -07:00
Alex Huang ecd9fad89c bug 10375: fixed a problem with build name 2011-06-22 16:43:53 -07:00
Alex Huang 252ca3f81a build better names 2011-06-22 16:34:14 -07:00
Alex Huang 47f114ad96 faster implementation of the log renderer 2011-06-22 16:34:14 -07:00
Sheng Yang cba8e16b61 Catch more exception 2011-06-22 14:46:15 -07:00
Sheng Yang 64b2293a50 Add more _selector.close()
Also add more log.
2011-06-22 14:24:12 -07:00
Alex Huang 573c054673 bug 10375: Added logging for exception causes 2011-06-22 11:51:35 -07:00
Alex Huang 0cf9d8b788 some connection logging was missing 2011-06-21 14:42:12 -07:00
Alex Huang 22d5eac400 new log4j 2011-06-21 10:15:21 -07:00
Sheng Yang da0379149a SSL: Fix loop in when BUFFER_UNDERFLOW
Add 10 times as limit, as well as add the missing count=-1 judgement.
2011-06-20 18:36:35 -07:00
Alex Huang e4f42d0e52 bug 10368: Print out the hash code of the db connection instead of the toString(). Changed the transaction in changeToStart() method to be of a shorter interval so it won't cause so many db connection exceptions 2011-06-20 11:30:23 -07:00
Alex Huang c9a58ba36e added connection logging to check for db connection losses 2011-06-19 06:31:33 -07:00