| Summary: | Improve handling of damaged / partial vmcores: add --zero_excluded and --minimal to 'crash_cmd' if certain crash failures occur | ||
|---|---|---|---|
| Product: | [Fedora] Fedora EPEL | Reporter: | Dave Wysochanski <dwysocha> |
| Component: | retrace-server | Assignee: | Dave Wysochanski <dwysocha> |
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
| Severity: | medium | Docs Contact: | |
| Priority: | low | ||
| Version: | el6 | CC: | abrt-devel-list, jberan, michal.toman |
| 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: | 2018-12-21 15:43:30 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
This probably should be tackled alongside https://bugzilla.redhat.com/show_bug.cgi?id=1232019 I am not sure how important this is but keeping it open for now. For now just set --minimal if we recognize the kernel and have a decent sized kernel log. https://github.com/abrt/retrace-server/pull/187 $ git tag --contains e27be24 1.19.0 |
Description of problem: Some vmcores are damaged but can be run with other options. For example, recently we received one vmcore and crash load showed the following: WARNING: /cores/retrace/tasks/776013158/crash/vmcore: This dumpfile is incomplete. This may cause the crash session to fail entirely, may cause commands to fail, or may result in unpredictable runtime behavior. NOTE: This dumpfile may be analyzed with the --zero_excluded command line option, in which case any read requests from missing pages will return zero-filled memory. I manually added the option via the following: $ echo -n "crash --zero_excluded" > /cores/retrace/tasks/776013158/crash_cmd We can then run other commands in the vmcore such as 'bt', etc. Version-Release number of selected component (if applicable): retrace-server-1.15-1.el6.noarch How reproducible: Once so far but depends on how many damaged / partial vmcores we get. Steps to Reproduce: TBD Actual results: "retrace-server-interact <taskid> crash" fails with crash exiting with an error, but the retrace-server task status == 'success' Expected results: "retrace-server-interact <taskid> crash" does not fail if some other option such as '--zero_excluded' would work. Additional info: This is not a high priority but it would help in some instances. We still get approximately 20% vmcores which fail in some way. There's some other options as well which may be useful, such as --no_kmem_cache, and of course, if all else fails, --minimal. Right now the way the code is structured for the 32-bit vmcore and the VMware --phys_base parameter should probably be refactored so we can add these other options. Also probably we need some patches so that running crash affects the task 'status' in some way, or there's a secondary status possibly. Will need to think about it and work on some patches to see what can be done. The one example we had occurred on a vmcore with kernel 3.10.0-327.13.1.el7.x86_64.debug