[Xen-users] Re: xm mem-set Domain-0 box went poof

[Brandon Lamb <brandonlamb@xxxxxxxxx>]

On Sun, Mar 14, 2010 at 2:03 AM, Brandon Lamb <brandonlamb@xxxxxxxxx> wrote:
> On Sun, Mar 14, 2010 at 1:49 AM, Brandon Lamb <brandonlamb@xxxxxxxxx> wrote:
>> On a fresh reboot
>>
>> free:
>>    total  used  Âfree shared buffers cached
>> Mem: 3371340 448156 2923184 Â Â Â0 Â Â5328 Â77840
>>
>> ----------
>>
>> top:
>> Mem: Â 3371340k total, Â 449044k used, Â2922296k free, Â Â 5344k buffers
>>
>> ----------
>>
>> xentop - 01:44:00 Â Xen 3.4.3-rc4-pre
>> 1 domains: 1 running, 0 blocked, 0 paused, 0 crashed, 0 dying, 0 shutdown
>> Mem: 4160828k total, 4022996k used, 137832k free  ÂCPUs: 4 @ 2308MHz
>> Â Â ÂNAME ÂSTATE Â CPU(sec) CPU(%) Â Â MEM(k) MEM(%) ÂMAXMEM(k)
>> MAXMEM(%) VCPUS NETS NETTX(k) NETRX(k) VBDS Â VBD_OO Â VBD_RD Â VBD_WR
>> SSID
>> ÂDomain-0 -----r    Â109  Â2.2  Â3965032  95.3  no limit    n/a  Â
>>  4
>>
>> ----------
>>
>> xm mem-set Domain-0 512
>>
>> xentop showed ~1G memory (26% memory), then right after this i tried
>>
>> xm mem-set Domain-0 192
>>
>> ... box died with caps and scroll lock flashing.
>>
>> I had dom0_mem=512, dom0-min-mem=192
>>
>> xen-3.4-testing-hg and xen/master kernel from jeremy's git repo
>
> Rebooted, this time did xm mem-set Domain-0 448, no crashes.
>
> xentop - 01:58:54 Â Xen 3.4.3-rc4-pre
> 1 domains: 1 running, 0 blocked, 0 paused, 0 crashed, 0 dying, 0 shutdown
> Mem: 4160828k total, 1113620k used, 3047208k free  ÂCPUs: 4 @ 2308MHz
> Â Â ÂNAME ÂSTATE Â CPU(sec) CPU(%) Â Â MEM(k) MEM(%) ÂMAXMEM(k)
> MAXMEM(%) VCPUS NETS NETTX(k) NETRX(k) VBDS Â VBD_OO Â VBD_RD Â VBD_WR
> SSID
> ÂDomain-0 -----r    Â123  Â1.8  Â1055656  25.4  no limit
> n/a   4  Â0    Â0    Â0  Â0    Â0    Â0    Â0  Â0
>
> So this look to me like I *should* have 3 gigs free, why its still
> saying that its using 1gig is beyond me. Anyhow, now I try to start my
> guest...
>
> xen1:/etc/xen# xm create syslog.cfg
> Using config file "./syslog.cfg".
> Error: (4, 'Out of memory', 'xc_dom_alloc_segment: segment ramdisk too
> large (0xef4f > 0x10000 - 0x26e0 pages)\n')
>
> xen1:/etc/xen# cat syslog.cfg
> kernel   Â= '/boot/vmlinuz-2.6.31.12'
> ramdisk   = '/boot/initrd.img-2.6.31.12'
> memory   Â= '256'
> #maxmem     = '384'
> root    Â= '/dev/xvda ro'
> disk    Â= ['phy:/dev/volumes/syslog,xvda,w']
> name    Â= 'syslog'
> vcpus      = 1
> vif     = [ 'ip=208.200.248.121, mac=00:16:3E:36:9C:15,
> bridge=xenbr0, model=e1000' ]
> vfb           = ['vnc=1, vnclisten=0.0.0.0, vncunused=1']
> serial     Â= 'pty'
> on_poweroff = 'destroy'
> on_reboot  = 'restart'
> on_crash  Â= 'restart'
>
> /dev/volumes/syslog is on lvm.
>
> I dont get it, I had this working fine on another box argh. The last
> thing I just did was remove all /boot files related to xen, dom0
> kernel etc, did make uninstall in xen-testing and then copied over
> tarballs of my working install, then ran the make install and copied
> my xend-config over. Why does this hate me so

Wow well just had a WTH moment. For s&g I set memory=512 in my
syslog.cfg... and it STARTED!

No more Error: (4, 'Out of memory', 'xc_dom_alloc_segment: segment
ramdisk too large (0xef4f > 0x10000 - 0x26e0 pages)\n')

I am totally scratching my head... must be something going on that i
dont know about

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users