View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001513 | GNUnet | ARM service | public | 2009-11-16 01:53 | 2010-02-15 14:37 |
| Reporter | Christian Grothoff | Assigned To | safey | ||
| Priority | normal | Severity | minor | Reproducibility | N/A |
| Status | closed | Resolution | fixed | ||
| Summary | 0001513: ARM should do an expontential back-off for service restarts after crashes during start-up | ||||
| Description | Currently, ARM restarts a managed process as soon as it dies. This can cause problems if the process consistently dies "immediately", i.e. because of a problem in the start-up sequence of the process. If this happens, the system load could go to 100% just because of re-spawning. In this case, ARM should detect that the process is consistently crashing (always after about N seconds) and delay restarting the process. If the process keeps crashing, the delay should be increased exponentially (times two, each time), until a certain maximum (say 1 day) is passed. Should the process run normally and then crash after some time >> N, the restart-delay should be reset to 1ms. Should the process crash much before the "expected" time N, the restart delay should probably just be kept (the same rule also applies if we have insufficient data to have a reasonable expectation value for N, say if the process has only died once before). | ||||
| Tags | No tags attached. | ||||
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2009-11-16 01:53 | Christian Grothoff | New Issue | |
| 2009-11-16 01:53 | Christian Grothoff | Status | new => assigned |
| 2009-11-16 01:53 | Christian Grothoff | Assigned To | => safey |
| 2010-02-13 00:07 | Christian Grothoff | Status | assigned => resolved |
| 2010-02-13 00:07 | Christian Grothoff | Resolution | open => fixed |
| 2010-02-15 14:37 | Christian Grothoff | Status | resolved => closed |
