Opened 13 years ago
Closed 9 years ago
#10745 closed defect (obsolete)
Guru Meditation 1103 (VINF_EM_DBG_HYPER_ASSERTION), error -1701 (VERR_MM_HYPER_NO_MEMORY)
| Reported by: | John Watson | Owned by: | |
|---|---|---|---|
| Component: | other | Version: | VirtualBox 4.1.18 |
| Keywords: | Guru Meditation 1103 | Cc: | |
| Guest type: | Linux | Host type: | Linux |
Description
Using VB 4.1.18. Host is OEL5.6 x86_64, guests are OEL5.7 x86. Not using 64bit guests because CPUs are the older AMD Opteron 248 that don't have hardware virtualization. I have told VB not to use HW virtualization, see "Hardw. virt.ext: off" in the attached showvminfo.txt file. I have three VMs: oel57net has 500M RAM, this never fails. It doesn't do much: acts as a DHCP and DNS and NFS server for the other two. oel57a and oel57b have 1500M RAM and fail regularly but unpredictably with a Guru Mediatation 1103. They never last more 15 minutes, they usually last at least 5 minutes. I had thought that the problem was to do with the software I am running (I'm building a Oracle 11.2 cluster) but I do not believe this: the failures sometimes occur before starting the clusterware daemons. I previously raised this issue on the forum when I was using VB release 4.1.12 this is the link tio the post https://forums.virtualbox.org/viewtopic.php?f=7&t=49184&p=223700&hilit=guru+meditation+1103#p223700 which includes more examples of the problem. Please tell me any other information I can provide, and thank you for any assistance.
Attachments (5)
Change History (14)
by , 13 years ago
| Attachment: | oel57b-2012-07-09-20-42-12.log added |
|---|
comment:1 by , 13 years ago
And to complete the example, another log file: no activity on the VM, booted it and left it alone. And by the way, sorry about my typo in the original post: host OS is OEL 5.5, not OEL 5.6.
by , 13 years ago
| Attachment: | oel57a-2012-07-10-10-37-11.log added |
|---|
by , 13 years ago
| Attachment: | oel57b-2012-07-10-10-48-03.log added |
|---|
comment:2 by , 13 years ago
As well as setting
Hardw. virt.ext: off
I have also tried
Page Fusion: off Hardw. virt.ext exclusive: off Nested Paging: off VT-x VPID: off
and there is no memory balloon. The 1500M VMs still fail unpredicatably after a few minutes.
comment:3 by , 13 years ago
This looks like you're running out of hyperheap memory. Could you try bumping the HyperHeap size from the default 1.25 MiB to say 3 MiB?
- Poweroff the VM.
- Enter the command:
VBoxManage setextradata <VMname> "VBoxInternal/MM/cbHyperHeap" "3145728"
- Start the VM.
Note: the value you specify is in bytes (here, 3145728 bytes = 3 MiB). If you continue to get this guru, try bumping this value even higher. The MMHyper code rounds this to 4096 bytes boundaries.
comment:4 by , 13 years ago
| Summary: | Guru Meditation 1103 (VINF_EM_DBG_HYPER_ASSERTION) → Guru Meditation 1103 (VINF_EM_DBG_HYPER_ASSERTION), error -1701 (VERR_MM_HYPER_NO_MEMORY) |
|---|
comment:5 by , 13 years ago
Thank you for your advice, ramshankar, it looks good. I am using the value you suggested, and the VMs have been running for a couple of hours no problem. Please close the ticket. I could only try it today, because I've been on a different job for the last two weeks.
comment:6 by , 13 years ago
| Resolution: | → worksforme |
|---|---|
| Status: | new → closed |
comment:7 by , 11 years ago
| Resolution: | worksforme |
|---|---|
| Status: | closed → reopened |
comment:8 by , 11 years ago
Better use
VBoxManage setextradata VM_NAME "VBoxInternal/MM/CanUseLargerHeap" 1
We should add some heuristics to VBox to set this option automatically.
comment:9 by , 9 years ago
| Resolution: | → obsolete |
|---|---|
| Status: | reopened → closed |
Please reopen if still relevant with a recent VirtualBox release.
VM logfile