Bug 819521
| Summary: | [plugin container] management of the Server Resource -> ProcessInfo mappings needs to be refactored | ||
|---|---|---|---|
| Product: | [Other] RHQ Project | Reporter: | Ian Springer <ian.springer> |
| Component: | Plugin Container, Performance | Assignee: | Nobody <nobody> |
| Status: | NEW --- | QA Contact: | |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 4.3 | CC: | hrupp, jshaughn |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | Type: | Bug | |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 620931 | ||
|
Description
Ian Springer
2012-05-07 13:20:49 UTC
per BZ triage ... crouch, loleary, foley Refreshing the process info on discovery scans is not enough... It opens up a possibility for unnoticed process restarts, where the process info would no longer correspond to the actual process the resource should be managing. The prominent plugin that needs always-up-to-date process info is Apache, because it actually needs the process info to be able to figure out the correct runtime configuration of the apache resource. As to "unnecessary discovery scans often occur" - the discovery scans only occur if the code in ResourceContext.getNativeProcess() establishes that the last known processInfo is no longer running - at that point new process scan is done and discovery is kicked off to see whether it can re-match the resource to the currently running process. > Refreshing the process info on discovery scans is not enough... > It opens up a > possibility for unnoticed process restarts, where > the process info would no longer correspond to the actual process > the resource should be managing. We would always verify the ProcessInfo is still running. If not, we would run a new discovery scan to get the new process. As part of this, I would enhance ProcessInfo to detect when the underlying process has died and mark the object as permanently defunct, and to make sure the process's name, command line, etc. has not changed. That way there would be much less chance of the ProcessInfo ending up representing some new process with the same pid as the original process. > The prominent plugin that needs always-up-to-date process info is > Apache, because it actually needs the process info to be able to > figure out the correct runtime configuration of the apache resource. It would still have this. The changes I'm suggesting would just make the underlying impl that returns the server's ProcessInfo more efficient, only doing discovery scans when absolutely necessary. > As to "unnecessary discovery scans often occur" - the discovery > scans only occur if the code in ResourceContext.getNativeProcess() > establishes that the last known processInfo is no longer running > - at that point new process scan is done and discovery is kicked > off to see whether it can re-match the resource to the currently > running process. As I recall, it looked to like a discovery scan would always be done the first time getNativeProcess() is called for a particular ResourceType. |