summaryrefslogtreecommitdiffstats
path: root/messagebus
Commit message (Collapse)AuthorAgeFilesLines
* Keep the spec final.Henning Baldersheim2019-05-284-21/+22
| | | | | | Create the address when needed in the async connect thread. Implement hash/equal/compareTo for Spec to avoid toString. Use Spec as key and avoid creating it every time.
* Reduce the length/cost of the test.Henning Baldersheim2019-05-271-3/+3
|
* Make the hot/cold path evident and readable.Henning Baldersheim2019-05-211-9/+17
|
* Add control over #connectionsHenning Baldersheim2019-05-201-1/+1
|
* Prepare for using multiple connections per specHenning Baldersheim2019-05-194-27/+110
|
* Revert "Revert "Wait until shutdown is completed.""Henning Baldersheim2019-05-161-1/+1
|
* Merge pull request #9438 from vespa-engine/havardpe/avoid-unit-test-hangHenning Baldersheim2019-05-161-2/+2
|\ | | | | band-aid to avoid blocking forever in transport shutdown
| * band-aid to avoid blocking forever in transport shutdownHåvard Pettersen2019-05-161-2/+2
| | | | | | | | | | make it more likely that messagebus transport thread(s) are started when start is called.
* | Revert "Wait until shutdown is completed."Harald Musum2019-05-161-1/+1
| |
* | Merge pull request #9435 from ↵Henning Baldersheim2019-05-161-1/+8
|\ \ | | | | | | | | | | | | vespa-engine/balder/ensure-we-do-not-drop-close-destroy-on-exception-or-create-some-that-we-do-not-destroy - Use double checked locking to ensure that we do not create Policies…
| * | - Use double checked locking to ensure that we do not create Policies that ↵Henning Baldersheim2019-05-161-1/+8
| |/ | | | | | | | | | | we forget to destroy. - Catch exceptions and close/destroy when necessary.
* | Merge pull request #9431 from vespa-engine/balder/2-mbus-transport-threadsHenning Baldersheim2019-05-161-2/+1
|\ \ | |/ |/| Use 2 threads for mbus transport.
| * Use 2 threads for mbus transport.Henning Baldersheim2019-05-161-2/+1
| |
* | Wait until shutdown is completed.Henning Baldersheim2019-05-161-1/+1
|/
* multi-threaded transport for JRTHåvard Pettersen2019-05-131-27/+2
|
* Simplify the supervisor responsibilityHenning Baldersheim2019-05-103-12/+7
|
* Explicitly require non-empty (or null) routeJon Bratseth2019-04-292-48/+38
|
* Test and error code categorization.Henning Baldersheim2019-04-292-2/+39
|
* Report fatal-mbus-errors-as server error.Henning Baldersheim2019-04-283-3/+16
|
* Merge pull request #9155 from vespa-engine/balder/update-lookup-interfaceHenning Baldersheim2019-04-235-19/+22
|\ | | | | Change interface from Mirror.Entry[] to List<Mirror.Entry> as you alr…
| * Change interface from Mirror.Entry[] to List<Mirror.Entry> as you already ↵Henning Baldersheim2019-04-225-19/+22
| | | | | | | | | | | | have a list. Avoid having to do an array copy that is not necessary.
* | Avoid '*' imports.Henning Baldersheim2019-04-221-1/+5
|/
* Cut number of network worker threads in half.Henning Baldersheim2019-04-221-4/+4
|
* Let the sender do all the packing too.Henning Baldersheim2019-04-201-1/+1
|
* Handle the reply in own thread too.Henning Baldersheim2019-04-201-26/+5
|
* Send the message in your own thread.Henning Baldersheim2019-04-201-31/+2
|
* Merge pull request #9134 from vespa-engine/balder/control-number-of-feed-threadsBjørn Christian Seime2019-04-152-6/+1
|\ | | | | Add control for number of feed threads.
| * Add control for number of feed threads.Henning Baldersheim2019-04-152-6/+1
| |
* | Revert "Balder/skip the messenger thread"Henning Baldersheim2019-04-143-24/+58
|/
* Add control for if you want to use mbus threas for compression/encoding or ↵Henning Baldersheim2019-04-142-1/+17
| | | | your own.
* Handle the reply in own thread too.Henning Baldersheim2019-04-131-26/+5
|
* Send the message in your own thread.Henning Baldersheim2019-04-131-31/+2
|
* Revert "Revert "Revert "Balder/send in own thread"""Henning Baldersheim2019-04-131-2/+31
|
* Remove some unused metrics code and dewire it.Henning Baldersheim2019-04-1311-204/+13
|
* GC unused metrics instead of guarding it.Henning Baldersheim2019-04-138-182/+0
|
* Use atomic and synchronized sections to do proper accounting of replies. And ↵Henning Baldersheim2019-04-133-4/+8
| | | | some geenral code health.
* Revert "Revert "Revert "Reply in own thread"""Henning Baldersheim2019-04-131-5/+26
|
* Revert "Revert "Reply in own thread""Henning Baldersheim2019-04-131-26/+5
|
* Revert "Reply in own thread"Henning Baldersheim2019-04-121-5/+26
|
* Merge pull request #9115 from ↵Henning Baldersheim2019-04-121-26/+5
|\ | | | | | | | | vespa-engine/balder/balder/reply-in-own-thread-only Reply in own thread
| * Reply in own threadHenning Baldersheim2019-04-121-26/+5
| |
* | Simplify with lambda.Henning Baldersheim2019-04-121-8/+2
| |
* | Take the time outside of the synchronized section.Henning Baldersheim2019-04-121-2/+4
|/
* A collection of code cleanup in messagebus. And a bonus of catching missing ↵Henning Baldersheim2019-04-1214-35/+25
| | | | shutdown of config subscription.
* Revert "Revert "Balder/send in own thread""Henning Baldersheim2019-04-121-31/+2
| | | | Will do the send in own thread and not dispatch to separate thread.
* Avoid universal importHenning Baldersheim2019-04-112-6/+32
|
* Update abi spec.Henning Baldersheim2019-04-111-1/+0
|
* Follow intellij advise.Henning Baldersheim2019-04-111-5/+5
|
* Check for retries at least every 10ms when no replies, as opposed to every ↵Henning Baldersheim2019-04-111-4/+4
| | | | 100ms.
* Eliminate some clang 7 warnings.Tor Egge2019-02-251-2/+2
|