Medium We can have a problem when
- Agent 1 takes Build A-1 from Plan A
- Agent 1 loses network connectivity to the server
- Agent 1 is marked as "going down" and goes off-line since the server could not receive the heartbeat. Lock on Plan A is released
- Build A-2 is triggered. Starts building on Local Agent
- Agent 1 comes back online, still with Build A-1.
At this point, things go wrong. Logs gets streamed in a funny way (since we assume that only one build is ever building), Build A-2 is saved with Build A-1. JIRAns have an example of this.
We should make sure if an agent has been marked as going offline, it is never allowed of come back (maybe check startup timestamp) when an Agent comes back online.
Needs further investigation and if the problem is too large to safely fix on branch, then we can defer it to fixing on trunk.
| Form Name | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
[BAM-2743] Remote agents may return even when already marked as dead
| Workflow | Original: Bamboo Workflow 2016 v1 - Restricted [ 1440725 ] | New: JAC Bug Workflow v3 [ 3382215 ] |
| Status | Original: Resolved [ 5 ] | New: Closed [ 6 ] |
| Workflow | Original: Bamboo Workflow 2016 v1 [ 1415735 ] | New: Bamboo Workflow 2016 v1 - Restricted [ 1440725 ] |
| Workflow | Original: Bamboo Workflow 2014 v2 [ 618801 ] | New: Bamboo Workflow 2016 v1 [ 1415735 ] |
| Workflow | Original: Bamboo Workflow 2014 [ 604983 ] | New: Bamboo Workflow 2014 v2 [ 618801 ] |
| Workflow | Original: Bamboo Workflow 2010 [ 202793 ] | New: Bamboo Workflow 2014 [ 604983 ] |
| Workflow | Original: reviewflow [ 165291 ] | New: Bamboo Workflow 2010 [ 202793 ] |
| Component/s | Original: Distributed Builds [ 12547 ] |
| Workflow | Original: jira [ 128144 ] | New: reviewflow [ 165291 ] |
| Fix Version/s | New: 2.0.6 [ 14025 ] | |
| Fix Version/s | Original: Next bug fix release [ 13890 ] | |
| Resolution | New: Fixed [ 1 ] | |
| Status | Original: In Progress [ 3 ] | New: Resolved [ 5 ] |
| Fix Version/s | Original: 2.1 [ 13140 ] |