From: Anthony Wright <anthony@overnetdata.com>
To: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: "xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>
Subject: Re: Memory allocation going seriously wonky on 4.1.1
Date: Thu, 06 Oct 2011 15:59:32 +0100 [thread overview]
Message-ID: <4E8DC254.7060001@overnetdata.com> (raw)
In-Reply-To: <20111003190917.GF15608@phenom.oracle.com>
[-- Attachment #1: Type: text/plain, Size: 2926 bytes --]
On 03/10/2011 20:09, Konrad Rzeszutek Wilk wrote:
> On Mon, Oct 03, 2011 at 02:45:36PM +0100, Anthony Wright wrote:
>> On 01/10/2011 13:07, Pasi Kärkkäinen wrote:
>>> On Fri, Sep 30, 2011 at 04:07:06PM +0100, Anthony Wright wrote:
>>>> I have a machine with 16GB of RAM running 32 bit xen 4.1.1 with Dom0
>>>> running a 3.0.4 linux kernel and 23 paravirtualized DomUs.
>>>>
>>> Hey,
>>>
>>> Why 32bit Xen? I think all (most?) hypervisor testing happens with 64bit Xen.
>>> Note that you can still have 32bit dom0 kernel with 64b hypervisor.
>> I'm going to investigate 64 bit xen on a 32 bit Dom0 soon, but as to the
>> 32 xen, I work quite a lot with older hardware and Xen since it does
>> paravirtualisation, and it's a really good way to re-use old hardware.
>>>> As the free ram gets to 2391 MB free the system behaves as if it's got
>>>> no more ram, and starts grabbing ram from Dom0, before getting even more
>>>> confused and failing to start DomUs and eventually crashing the whole
>>>> machine. I have attached a tarball with the output of xl dmesg, xl info
>>>> & xl list at various points.
>>>>
>>> - Do you use dom0_mem= parameter for Xen?
>>> - Did you disable dom0 ballooning?
>>>
>>> http://wiki.xen.org/xenwiki/XenBestPractices
>> I was already using the dom0_mem= parameter on the xen hypervisor, but I
>> disabled dom0 ballooning (enable-dom0-ballooning no), and just for good
>> measure I also set (dom0-min-mem 2000). This had no effect, when I
>> increase the 10-4 memory requirements, on restart the Dom0 memory is
>> reduced to 1713MB and xl info reports 2391MB free.
>>>> Stage 1
>>>> Everything is fine, Dom0 has 2000MB according to xl list, and 'xl info'
>>>> says there is 2405MB free.
>>> - How much memory did dom0 initially have?
>>> - Did you use mem= parameter for dom0 kernel (vmlinuz) ?
>>>
>>> See: http://wiki.xen.org/xenwiki/Linux_30_bugs
>> dom0 starts with 2000MB.
>> I didn't use mem= on the linux kernel - I thought it wasn't neccessary.
> It is if your kernel does not have these two patches:
>
> xen: use maximum reservation to limit amount of usable RAM
> xen/e820: if there is no dom0_mem=, don't tweak extra_pages.
>
> .. oh and you also need to update the hypervisor to include:
> "
> For dom0, this requires at patch to Xen ('x86: use 'dom0_mem' to limit
> the number of pages for dom0') (c/s 23790)
> "
>
> Grrr.. Let me update the 3.0 bug-list
>From reading around some of the bug reports, I thought this was a
problem caused by the fact that I have 16GB of ram. As a result I
reduced the amount of physical ram in the machine down to 8GB, but still
got the same effect (with the two kernel patches, but not with the xen
patch as I can't find it). I've attached a log of xl info & xl list as
the domU that trips the machine up is started (10-5). Dom0 is supposed
to have 700MB of ram (dom0_mem=500M), dom0-min-mem is 700 and
enable-dom0-ballooning is no.
[-- Attachment #2: dom-mem.log --]
[-- Type: text/plain, Size: 19241 bytes --]
AET-Box-4:~ # date ; xl info ; xl list
Thu Oct 6 14:53:19 UTC 2011
host : AET-Box-4
release : 3.0.4
version : #1 SMP Mon Oct 3 05:22:53 GMT 2011
machine : i686
nr_cpus : 16
nr_nodes : 2
cores_per_socket : 4
threads_per_core : 2
cpu_mhz : 2533
hw_caps : bfebfbff:2c100000:00000000:00003f40:029ee3ff:00000000:00000001:00000000
virt_caps : hvm hvm_directio
total_memory : 8181
free_memory : 1491
free_cpus : 0
xen_major : 4
xen_minor : 1
xen_extra : .1
xen_caps : xen-3.0-x86_32p hvm-3.0-x86_32 hvm-3.0-x86_32p
xen_scheduler : credit
xen_pagesize : 4096
platform_params : virt_start=0xf5800000
xen_changeset : unavailable
xen_commandline : dom0_mem=700M sched=credit
cc_compiler : gcc version 4.4.3 (GCC)
cc_compile_by :
cc_compile_domain : [unknown]
cc_compile_date : Wed Sep 21 08:25:36 GMT 2011
xend_config_format : 4
Name ID Mem VCPUs State Time(s)
Domain-0 0 700 8 r----- 64.3
1-15 1 1300 8 -b---- 9.3
22-5 2 400 2 -b---- 8.9
7-5 3 400 2 -b---- 8.5
8-4 4 200 1 -b---- 3.3
16-5 5 400 2 -b---- 6.2
12-2 6 200 1 -b---- 4.3
21-2 7 200 1 -b---- 4.1
4-5 8 400 2 -b---- 6.0
15-2 9 200 1 -b---- 4.3
5-4 10 200 1 -b---- 2.7
19-5 11 400 2 -b---- 4.3
17-4 12 200 1 -b---- 2.5
13-5 13 400 2 -b---- 1.4
20-4 14 200 1 -b---- 2.8
11-4 15 200 1 -b---- 2.5
14-4 16 200 1 -b---- 1.3
23-6 17 200 1 -b---- 1.1
6-2 18 200 1 --p--- 0.0
AET-Box-4:~ # date ; xl info ; xl list
Thu Oct 6 14:53:23 UTC 2011
host : AET-Box-4
release : 3.0.4
version : #1 SMP Mon Oct 3 05:22:53 GMT 2011
machine : i686
nr_cpus : 16
nr_nodes : 2
cores_per_socket : 4
threads_per_core : 2
cpu_mhz : 2533
hw_caps : bfebfbff:2c100000:00000000:00003f40:029ee3ff:00000000:00000001:00000000
virt_caps : hvm hvm_directio
total_memory : 8181
free_memory : 1635
free_cpus : 0
xen_major : 4
xen_minor : 1
xen_extra : .1
xen_caps : xen-3.0-x86_32p hvm-3.0-x86_32 hvm-3.0-x86_32p
xen_scheduler : credit
xen_pagesize : 4096
platform_params : virt_start=0xf5800000
xen_changeset : unavailable
xen_commandline : dom0_mem=700M sched=credit
cc_compiler : gcc version 4.4.3 (GCC)
cc_compile_by :
cc_compile_domain : [unknown]
cc_compile_date : Wed Sep 21 08:25:36 GMT 2011
xend_config_format : 4
Name ID Mem VCPUs State Time(s)
Domain-0 0 555 8 r----- 65.3
1-15 1 1300 8 -b---- 9.5
22-5 2 400 2 -b---- 8.9
7-5 3 400 2 -b---- 8.5
8-4 4 200 1 -b---- 3.4
16-5 5 400 2 -b---- 6.5
12-2 6 200 1 -b---- 4.3
21-2 7 200 1 -b---- 4.2
4-5 8 400 2 -b---- 6.1
15-2 9 200 1 -b---- 4.3
5-4 10 200 1 -b---- 2.8
19-5 11 400 2 -b---- 4.8
17-4 12 200 1 -b---- 2.8
13-5 13 400 2 -b---- 1.4
20-4 14 200 1 -b---- 2.8
11-4 15 200 1 -b---- 2.6
14-4 16 200 1 -b---- 1.5
23-6 17 200 1 -b---- 1.2
6-2 18 200 1 -b---- 0.8
(null) 19 0 0 --p--- 0.0
AET-Box-4:~ # date ; xl info ; xl list
Thu Oct 6 14:53:28 UTC 2011
host : AET-Box-4
release : 3.0.4
version : #1 SMP Mon Oct 3 05:22:53 GMT 2011
machine : i686
nr_cpus : 16
nr_nodes : 2
cores_per_socket : 4
threads_per_core : 2
cpu_mhz : 2533
hw_caps : bfebfbff:2c100000:00000000:00003f40:029ee3ff:00000000:00000001:00000000
virt_caps : hvm hvm_directio
total_memory : 8181
free_memory : 1635
free_cpus : 0
xen_major : 4
xen_minor : 1
xen_extra : .1
xen_caps : xen-3.0-x86_32p hvm-3.0-x86_32 hvm-3.0-x86_32p
xen_scheduler : credit
xen_pagesize : 4096
platform_params : virt_start=0xf5800000
xen_changeset : unavailable
xen_commandline : dom0_mem=700M sched=credit
cc_compiler : gcc version 4.4.3 (GCC)
cc_compile_by :
cc_compile_domain : [unknown]
cc_compile_date : Wed Sep 21 08:25:36 GMT 2011
xend_config_format : 4
Name ID Mem VCPUs State Time(s)
Domain-0 0 555 8 r----- 66.4
1-15 1 1300 8 -b---- 9.6
22-5 2 400 2 -b---- 8.9
7-5 3 400 2 -b---- 8.5
8-4 4 200 1 -b---- 3.4
16-5 5 400 2 -b---- 7.0
12-2 6 200 1 -b---- 4.4
21-2 7 200 1 -b---- 4.2
4-5 8 400 2 -b---- 6.5
15-2 9 200 1 -b---- 4.3
5-4 10 200 1 -b---- 2.8
19-5 11 400 2 -b---- 5.5
17-4 12 200 1 -b---- 2.8
13-5 13 400 2 -b---- 1.5
20-4 14 200 1 -b---- 3.0
11-4 15 200 1 -b---- 2.7
14-4 16 200 1 -b---- 2.2
23-6 17 200 1 -b---- 2.2
6-2 18 200 1 -b---- 0.8
10-5 19 0 0 --p--- 0.0
AET-Box-4:~ # date ; xl info ; xl list
Thu Oct 6 14:53:30 UTC 2011
host : AET-Box-4
release : 3.0.4
version : #1 SMP Mon Oct 3 05:22:53 GMT 2011
machine : i686
nr_cpus : 16
nr_nodes : 2
cores_per_socket : 4
threads_per_core : 2
cpu_mhz : 2533
hw_caps : bfebfbff:2c100000:00000000:00003f40:029ee3ff:00000000:00000001:00000000
virt_caps : hvm hvm_directio
total_memory : 8181
free_memory : 1235
free_cpus : 0
xen_major : 4
xen_minor : 1
xen_extra : .1
xen_caps : xen-3.0-x86_32p hvm-3.0-x86_32 hvm-3.0-x86_32p
xen_scheduler : credit
xen_pagesize : 4096
platform_params : virt_start=0xf5800000
xen_changeset : unavailable
xen_commandline : dom0_mem=700M sched=credit
cc_compiler : gcc version 4.4.3 (GCC)
cc_compile_by :
cc_compile_domain : [unknown]
cc_compile_date : Wed Sep 21 08:25:36 GMT 2011
xend_config_format : 4
Name ID Mem VCPUs State Time(s)
Domain-0 0 555 8 r----- 67.2
1-15 1 1300 8 -b---- 9.7
22-5 2 400 2 -b---- 8.9
7-5 3 400 2 -b---- 8.5
8-4 4 200 1 -b---- 3.4
16-5 5 400 2 r----- 7.5
12-2 6 200 1 -b---- 4.4
21-2 7 200 1 -b---- 4.2
4-5 8 400 2 r----- 6.9
15-2 9 200 1 -b---- 4.3
5-4 10 200 1 -b---- 2.8
19-5 11 400 2 -b---- 5.7
17-4 12 200 1 -b---- 2.8
13-5 13 400 2 -b---- 1.5
20-4 14 200 1 -b---- 3.0
11-4 15 200 1 -b---- 2.8
14-4 16 200 1 -b---- 2.7
23-6 17 200 1 -b---- 2.7
6-2 18 200 1 -b---- 0.9
10-5 19 400 1 --p--- 0.0
[-- Attachment #3: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
next prev parent reply other threads:[~2011-10-06 14:59 UTC|newest] Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top 2011-09-30 15:07 Memory allocation going seriously wonky on 4.1.1 Anthony Wright 2011-10-01 12:07 ` Pasi Kärkkäinen 2011-10-03 13:45 ` Anthony Wright 2011-10-03 19:09 ` Konrad Rzeszutek Wilk 2011-10-06 12:57 ` Anthony Wright 2011-10-06 15:48 ` Konrad Rzeszutek Wilk 2011-10-06 15:51 ` Anthony Wright 2011-10-06 14:59 ` Anthony Wright [this message] 2011-10-03 13:55 ` Anthony Wright 2011-10-03 19:10 ` Konrad Rzeszutek Wilk 2011-10-06 14:05 ` Anthony Wright 2011-10-06 15:48 ` Konrad Rzeszutek Wilk 2011-10-03 19:04 ` Konrad Rzeszutek Wilk
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4E8DC254.7060001@overnetdata.com \
--to=anthony@overnetdata.com \
--cc=konrad.wilk@oracle.com \
--cc=xen-devel@lists.xensource.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.