1 KDUMP TESTS AUTOMATION SUITE
2 ----------------------------
4 The kdump test automation suite helps run the kdump tests and report
5 results. The testscripts cycle through a series of crash
6 scenarios. Each test cycle does the following:
8 1. Sets up a crash scenario.
10 3. Kdump kernel boots and saves a vmcore.
11 4. System reboots to 1st kernel.
12 5. vmcore is validated and results are saved.
13 6. After a 1 to 2 minute delay, the next crash scenario is setup and
16 The scripts make use of the crasher module for basic testing of kdump
17 and the new Linux Kernel Dump Test Module (LKDTM) for more involved
18 testing. LKDTM makes use of the kprobes infrastructure for inserting
19 crashpoints into the kernel at run-time. Thus the kernel need not be
20 patched and rebuilt.This script also tests kdump dumping on different
21 destinations like ext3, raw and network dump.
23 KDUMP TEST INSTRUCTION
24 ----------------------
26 Follow the steps to setup kdump test automation suite.
28 The tests are written for SuSE Linux Enterprise Server 10 (and onward
29 releases), OpenSUSE, Fedora, Debian, as well as RedHat Enterprise Linux
30 5. Since KDUMP is supported by the above mentioned distro's the tests
31 were written and tested on them. Contribution towards supporting more
32 distributions are welcome.
34 1. Install these additional packages:
36 For SLES10 or OpenSUSE Distro :
41 * zlib-64bit-<xxx> (ppc64 only)
42 * expect (if dump to a network destination)
44 For RHEL5 or Fedora distro :
48 * kernel-debuginfo rpm (if using crash to verify vmcore)
49 * kernel-kdump ( only for ppc64 )
50 * expect (if dump to a network destination)
52 2. Make sure the partition where the tests are running has space for
53 the tests results and one vmcore file (size of physical memory).
55 3. Modify configuration file runkdump.conf or copy a existing one from
58 4. Run "make". Carefully check for any errors.
60 Few Important points to remember:
62 * If you need to stop the tests before all tests have run, run "crontab
63 -r" and "killall runkdump.sh" within 1 minute after the 1st kernel
64 reboots. Then, if you'd like to carry on tests from the point on, run
65 "crontab kdump.cron", and "./runkdump.sh". If you'd like to start tests
66 from the beginning, modify the configuration file, and set
67 "REBOOT=0". Then, "./runkdump.sh"
69 * A failure is likely to occur when booting the kdump kernel. If this
70 happens, you'll need to manually reset the system so it reboots back to
71 the 1st kernel and continues on to the next test. For this reason, it's
72 best to monitor the tests from a console. If possible, setup a serial
73 console (not a must, any type of console setup will do). If using
74 minicom, enable saving of kernel messages displayed on minicom into a
75 file, by pressing ctrl+a+l on the console. Else, when it is observed
76 that the kdump kernel has failed to boot, manually copy the boot message
77 into a file to enable the debugging the cause of the hang.
79 * The results are saved in
80 <kdump-test-dir>/log/<hostname>.<date-time>. The "status" file in that
81 directory shows where you are in the test run. When the "Test run
82 complete" entry appears in that file, you're done. Verbose log can be
83 found at /tmp/kdump-<date>.log.
85 * The test machine would be unavailabe for any other work during the
86 period of the test run.
88 * System may hang if incorrect partition information is provided for
89 dumping, like specifying a partition which does not exist, specifying a
90 parition label which does not exist. This is not ltp kdump bug.