delta/rabbitmq-server.git, branch bug25640 hg.rabbitmq.com: Obsolete (use rabbitmq-server-git) prevent invalid state transition 2013-06-22T00:12:20+00:00 Matthias Radestock matthias@rabbitmq.com 2013-06-22T00:12:20+00:00 dfade0b0a301eaa921d5a1896b4d409d10e9c25f from 'dormant' to 'active' w/o registration This was very hard to trigger, but nevertheless possible. 
from 'dormant' to 'active' w/o registration

This was very hard to trigger, but nevertheless possible.
Make it clearer what I meant. 2013-06-21T14:48:50+00:00 Simon MacMullen simon@rabbitmq.com 2013-06-21T14:48:50+00:00 0371265eb754b0e7b364d1a5d9e3c7624f59f91f 






Merged bug25638 into stable
2013-06-21T14:41:56+00:00

Emile Joubert
emile@rabbitmq.com

2013-06-21T14:41:56+00:00

240e64eaf4a1f3ff5b190df5abf1008a6ebffa7c












Merged bug25615 into stable
2013-06-21T14:00:03+00:00

Emile Joubert
emile@rabbitmq.com

2013-06-21T14:00:03+00:00

175be28d678dcd6cb8d13afbec632237f76c1144












Add a TODO for if we are ever unbusy enough to look at removing this code.
2013-06-21T10:28:32+00:00

Simon MacMullen
simon@rabbitmq.com

2013-06-21T10:28:32+00:00

85953f05729171ae2a6a3285ca0bb63e353d6654












Don't persist gm_pids; by definition when all we have is the rabbit_durable_queue record the cluster has gone down completely so there can't be any pids. Fixes gm_pids leak. Since we're here, fix a similar but less severe leak in sync_slave_pids. (Less severe because rabbit_mirror_queue_misc:store_updated_slaves/1 has the effect of cleaning it up periodically. But it's still wrong.)
2013-06-21T10:26:15+00:00

Simon MacMullen
simon@rabbitmq.com

2013-06-21T10:26:15+00:00

0c3f3e71f9d41361f8afcd903c54db18a5d8385e












Merge bug25636 into stable
2013-06-20T17:35:10+00:00

Tim Watson
tim@rabbitmq.com

2013-06-20T17:35:10+00:00

4708a53c6219e8ab77a49ab18b9219fe0b6484e3












Start asn1 too.
2013-06-20T15:50:55+00:00

Simon MacMullen
simon@rabbitmq.com

2013-06-20T15:50:55+00:00

08f8f95412c45cf98a7a823e3c1eb30c504c4b68












Umm, be less than 100% wrong.
2013-06-20T14:25:25+00:00

Simon MacMullen
simon@rabbitmq.com

2013-06-20T14:25:25+00:00

b72670200638d5848aad710bcf3396eeee9c7fc0












Do not "recover" queues if there is already a rabbit_queue record for them. If we are restarting quickly it could be the case that we have a slave alive that has not failed over yet. Don't usurp it.
2013-06-20T13:48:31+00:00

Simon MacMullen
simon@rabbitmq.com

2013-06-20T13:48:31+00:00

92edfb501c0bd34f58e6ce9a8a4f7c9d13ddacd7