Xen PV wrong memory amount

Zero

Hello all,

I've bought a Xen PV VPS with 512 MB of RAM and 512 MB of Swap, but when I execute free -m I see:

total used free shared buffers cached
Mem: 497 46 451 0 1 22
-/+ buffers/cache: 21 475
Swap: 511 0 511

Is reserved by the kernel or what?
I haven't had this problem with others Xen VPS.

But if I reload the VPS using the CentOS 5 i386 template I get 512 MB of RAM...
The CentOS 5 template is the only that "works".

Is something that I can fix?

netomx

What is the problem?

Zero

Where are the remaining 15 MB of memory?

MrAndroid

@netomx said: What is the problem?

Im guessing its because the VPS shows 497MB of RAM, when he should have 512MB.

KuJoe

I believe that is caused by PyGrub, if you use a 64bit OS you should have more available RAM than a 32bit OS. I didn't test this extensively but I think that's the reason.

MrAndroid

@KuJoe said: I believe that is caused by PyGrub, if you use a 64bit OS you should have more available RAM than a 32bit OS. I didn't test this extensively but I think that's the reason.

You can also use PAE to address the full memory amount.

Zero

I get even less memory with CentOS 6 x86_64

total used free shared buffers cached
Mem: 489 67 422 0 1 22
-/+ buffers/cache: 43 446
Swap: 511 0 511

netomx

Can I ask why are you using 64 bits on a 512MB VPS ?

KuJoe

@netomx said: Can I ask why are you using 64 bits on a 512MB VPS ?

No PyGrub should result in more available memory. Guess I was wrong.

dmmcintyre3

The kernel uses some of your RAM.

efball

I have an 80MB xen box and when I switch from Debian 5 to Debian 6 with pygrub (both 32 bit) the RAM dropped to 73MB. pygrub uses more memory.

MrAndroid

@efball said: I have an 80MB xen box and when I switch from Debian 5 to Debian 6 with pygrub (both 32 bit) the RAM dropped to 73MB. pygrub uses more memory.

Sometimes the RAM is reserved by certain things.

Give the output of cat /proc/iomem

That will tell us what's exactly using your memory.

nabo

@Zero said: I get even less memory with CentOS 6 x86_64

total used free shared buffers cached
Mem: 489 67 422 0 1 22

512MB = 488MiB :-)

MrAndroid

@nabo said: 512MB = 488MiB. :-)

Wouldn't 512 Megabytes be 4096 Megabits?

nabo

@Daniel said: Wouldn't 512 Megabytes be 4096 Megabits?

MB = Megabyte
MiB = Mebibyte
Mb = Megabit

netomx

Why binary?

efball

@Daniel said: @efball said: I have an 80MB xen box and when I switch from Debian 5 to Debian 6 with pygrub (both 32 bit) the RAM dropped to 73MB. pygrub uses more memory.

Sometimes the RAM is reserved by certain things.
Give the output of cat /proc/iomem
That will tell us what's exactly using your memory.

cat /proc/iomem
00000000-0009ffff : System RAM
000a0000-000fffff : reserved
  000f0000-000fffff : System ROM
00100000-04ffffff : System RAM
  01000000-012841ca : Kernel code
  012841cb-013dfeb3 : Kernel data
  01447000-0150356b : Kernel bss

This is from the 80MB box that shows 73MB total with pygrub. It doesn't mean anything to me.

Francisco

it's likely the 'video card' eating some of it.

Francisco

MrAndroid

@efball it simply shows what memory address's are being used for.

efball

Here's some more data. All are using Debian 6:

32 bit, xen, pygrub, 128MB RAM, free shows 121MB = 7MB used
64 bit, xen, pygrub, 128MB RAM, free shows 118MB = 10MB used
32 bit, KVM, 128MB RAM, free shows 121MB = 7MB used

and here's the /proc/iomem from the KVM box:

cat /proc/iomem 
00000000-0009f7ff : System RAM
0009f800-0009ffff : reserved
000a0000-000bffff : Video RAM area
000c0000-000c7fff : Video ROM
000f0000-000fffff : reserved
  000f0000-000fffff : System ROM
00100000-07ffcfff : System RAM
  01000000-01272a56 : Kernel code
  01272a57-013be4c3 : Kernel data
  01422000-014ccd2b : Kernel bss
07ffd000-07ffffff : reserved
f0000000-f00000ff : 0000:00:02.0
  f0000000-f00000ff : 8139cp
f0010000-f001ffff : 0000:00:02.0
fec00000-fec00fff : IOAPIC 0
fee00000-fee00fff : Local APIC
fffbc000-ffffffff : reserved

So the pygrub xen boxes aren't any worse than KVM.

budingyun

cat /proc/iomem
00000000-00000fff : reserved
00001000-0009ffff : System RAM
000a0000-000fffff : reserved
  000f0000-000fffff : System ROM
00100000-7fffffff : System RAM
  01000000-01508274 : Kernel code
  01508275-01c07b2f : Kernel data
  01d4e000-02012023 : Kernel bss
  05000000-0d0fffff : Crash kernel

My Xen PV VPS have 2GB RAM but showing only 1.83GB RAM. What seems to be the problem?

AnthonySmith

Xen does not always report memory correctly through free -m

Run: cat /proc/meminfo

Near the bottom you should see: driectmap 4k, that is how much ram your VPS can actually use, if you see the wrong number in there then you have a problem but I suspect everything will be in order

budingyun

@AnthonySmith said: Run: cat /proc/meminfo

cat /proc/meminfo
MemTotal:        1914244 kB
MemFree:         1491196 kB
Buffers:            5384 kB
Cached:           222816 kB
SwapCached:            0 kB
Active:           152028 kB
Inactive:         200148 kB
Active(anon):     123992 kB
Inactive(anon):      320 kB
Active(file):      28036 kB
Inactive(file):   199828 kB
Unevictable:           0 kB
Mlocked:               0 kB
SwapTotal:       4194296 kB
SwapFree:        4194296 kB
Dirty:               172 kB
Writeback:             0 kB
AnonPages:        124008 kB
Mapped:            16896 kB
Shmem:               328 kB
Slab:              30664 kB
SReclaimable:      10540 kB
SUnreclaim:        20124 kB
KernelStack:        1088 kB
PageTables:         5648 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:     5151416 kB
Committed_AS:     764152 kB
VmallocTotal:   34359738367 kB
VmallocUsed:       10676 kB
VmallocChunk:   34359705620 kB
HardwareCorrupted:     0 kB
AnonHugePages:         0 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:     2097152 kB
DirectMap2M:           0 kB

Thanks.

dmmcintyre3

@AnthonySmith said: Run: cat /proc/meminfo

Near the bottom you should see: driectmap 4k, that is how much ram your VPS can actually use

On a KVM with 128mb memory:

DirectMap4k:        8180 kB
DirectMap2M:      122880 kB

I added the 2 numbers up and got 127.99mb.

AnthonySmith

@dmmcintyre3 not sure what that means in KVM as you dont get a dedicated memory allocation, you dont need to do anything special in KVM to vastly oversell ram so I guess what I said is only applicable on XenPV.

At a guess I would say that the total you see is what you have in use as that is all the host node needs to allocate to you at the time?

budingyun

dmesg | grep Memory
Memory: 1875180k/2097152k available (5152k kernel code, 388k absent, 221584k reserved, 7166k data, 1260k init)

Just wondering, it actually reserved for what?