ISST-LTE:pVM:roselp4:ubuntu 17.04: kdump failed after memory DLPAR

[bugproxy]

---Problem Description---
After a memory DLPAR removal, kdump doesn't work:

         Starting Kernel crash dump capture service...
[ 67.714593] kdump-tools[3850]: Starting kdump-tools: * running makedumpfile -c -d 31 /proc/vmcore /var/crash/201702230005/dump-incomplete
Copying data : [ 2.1 %] -/usr/sbin/kdump-config: line 639: 3897 Bus error makedumpfile $MAKEDUMP_ARGS $vmcore_file $KDUMP_CORETEMP
[ 72.314140] kdump-tools[3850]: * kdump-tools: makedumpfile failed, falling back to 'cp'
[ 73.693881] kdump-tools[3850]: cp: error reading '/proc/vmcore': Bad address
[ 73.704152] kdump-tools[3850]: * kdump-tools: failed to save vmcore in /var/crash/201702230005
[ 73.823643] kdump-tools[3850]: * running makedumpfile --dump-dmesg /proc/vmcore /var/crash/201702230005/dmesg.201702230005
[ 73.973813] kdump-tools[3850]: The kernel version is not supported.
[ 73.974078] kdump-tools[3850]: The makedumpfile operation may be incomplete.
[ 73.983506] kdump-tools[3850]: The dmesg log is saved to /var/crash/201702230005/dmesg.201702230005.
[ 73.983752] kdump-tools[3850]: makedumpfile Completed.
[ 73.983998] kdump-tools[3850]: * kdump-tools: saved dmesg content in /var/crash/201702230005
[ 74.104555] kdump-tools[3850]: Thu, 23 Feb 2017 00:05:15 -0600
[ 74.233502] kdump-tools[3850]: Failed to read reboot parameter file: No such file or directory
[ 74.233782] kdump-tools[3850]: Rebooting.
[ 86.629777] reboot: Restarting system

The kdump service should be restarted after the memory DLPAR operation.

C
---uname output---
Linux roselp4 4.10.0-8-generic #10-Ubuntu SMP Mon Feb 13 14:00:06 UTC 2017 ppc64le ppc64le ppc64le GNU/Linux

Machine Type = lpar

---Debugger---
A debugger is not configured

---Steps to Reproduce---
1. config kdump on roselp4
2. do a memory DLPAR removal operation
3. trigger kdump

[Manoj Iyer (manjo)]

/proc/vmcore is generated by kdump support in the kernel. This might not be a bug, but a case of missing support for regenerating /proc/vmcore entries after DLPAR memory removal.

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2017-03-21 13:08 EDT-------
(In reply to comment #5)
> /proc/vmcore is generated by kdump support in the kernel. This might not be
> a bug, but a case of missing support for regenerating /proc/vmcore entries
> after DLPAR memory removal.

True. kdump needs to be reloaded after a DLPAR operation to
regenerate /proc/vmcore entries. So, kdump-tools.service needs
to be restarted after a DLPAR operation.
Simialr bug raised against 16.04.2 - https://bugs.launchpad.net/bugs/1655280

Thanks
Hari

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2017-06-29 03:16 EDT-------
Any update

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2017-07-20 02:33 EDT-------
Any update

[Nish Aravamudan (nacc)]

How is this supposed to work generally. On DLPAR, kdump needs to be reloaded -- does that the mean the kdump.service should be watching for dlpar events?

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2017-07-31 13:49 EDT-------
(In reply to comment #10)
> How is this supposed to work generally. On DLPAR, kdump needs to be reloaded
> -- does that the mean the kdump.service should be watching for dlpar events?

Yes.. Probably, a udev event to look for dlpar events and
try-restart kdump-tools.service

Thanks
Hari

[Joseph Salisbury (jsalisbury)]

Hi Nish,

Do you thing bug 1655280 is a duplicate of this bug?

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2017-12-05 02:56 EDT-------
Any update on this bug

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2018-04-18 00:16 EDT-------
The fix mentioned in Launchpad bug 1655280 resolves this issue as well..

Thanks
Hari

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2018-04-18 00:21 EDT-------
(In reply to comment #16)
> The fix mentioned in Launchpad bug 1655280 resolves this issue as well..

Basically, a udev rule file that looks like this:

SUBSYSTEM=="memory", ACTION=="online", PROGRAM="/bin/systemctl try-restart kdump-tools.service"
SUBSYSTEM=="memory", ACTION=="offline", PROGRAM="/bin/systemctl try-restart kdump-tools.service"
SUBSYSTEM=="cpu", ACTION=="add", PROGRAM="/bin/systemctl try-restart kdump-tools.service"
SUBSYSTEM=="cpu", ACTION=="remove", PROGRAM="/bin/systemctl try-restart kdump-tools.service"

put in "/lib/udev/rules.d" dir as part of the kdump-tools package to avoid the need to
manually reload kdump..

Thanks
Hari