Posts by admin.

OpenSolaris 2009.06 to OpenSolaris preview 2010.02 (build 124)

Create a new Boot Environment:
pfexec beadm create devel

Mount the new Boot Environment:
pfexec mkdir /mnt/devel
pfexec beadm mount devel /mnt/devel

Use the dev publisher:
pfexec pkg -R /mnt/devel set-publisher -O

Update to the latest bits:
pfexec pkg -R /mnt/devel image-update -v

If everything went fine (after you read the Release Notes), activate the new BE:
pfexec beadm activate devel

pfexec shutdown -y -g1 -i6

Did this remotely. After reboot, system was up and running, including all xVM domU autostarted.

libima should get over pkginfo love

From Solaris Patch 119091-34

Problem Description:

6801126 libima should get over pkginfo love

I sure hope libima got over pkginfo love!

$20 Xen VPS Gandi vs. Linode vs. Slicehost

Gandi Linode Slicehost
Plan 1 Share Linode 360 256 slice
Price 14.35 EUR
(~20 USD)
19.95 USD 20 USD
RAM 256MB 360MB 256MB
Storage 8GB
(3 OS + 5 Data)
16GB 10GB
Bandwidth 5 Mbit 200GB 100GB
My VPS Location Paris, France
Dallas, TX
St. Louis, MI
My VPS CPU Quad-Core AMD Opteron(tm) Processor 8346 HE Intel(R) Xeon(R) CPU L5420 @ 2.50GHz Dual Core AMD Opteron(tm) Processor 270
My VPS vCPUs 1 4 4
My VPS OS CentOS 5.3 CentOS 5.3 CentOS 5.3
My VPS Buffered disk reads 10.48 MB/sec 73.77 MB/sec 60.86 MB/sec
My VPS OGR-NG Benchmark 23,608,069 nodes/sec 40,910,174 nodes/sec 17,593,557 nodes/sec
My VPS RC5-72 Benchmark 7,703,097 keys/sec 9,297,850 keys/sec 8,585,510 keys/sec

Linux /proc/cpuinfo flags

fpu – Onboard FPU
vme – Virtual Mode Extensions
de – Debugging Extensions
pse – Page Size Extensions
tsc – Time Stamp Counter
msr – Model-Specific Registers
pae – Physical Address Extensions
mce – Machine Check Architecture
cx8 – CMPXCHG8 instruction
apic – Onboard APIC
mtrr – Memory Type Range Registers
pge – Page Global Enable
mca – Machine Check Architecture
cmov – CMOV instructions (plus FCMOVcc, FCOMI with FPU)
pat – Page Attribute Table
pse36 – 36-bit PSEs
pn – Processor serial number
clflush – CLFLUSH instruction
dts – Debug Store
acpi – ACPI via MSR
mmx – Multimedia Extensions
sse – SSE
sse2 – SSE2
ss – CPU self snoop
ht – Hyper-Threading
tm – Automatic clock control
ia64 – IA-64 processor
pbe – Pending Break Enable

mp – MP Capable
nx – Execute Disable
mmxext – AMD MMX extensions
fxsr_opt – FXSAVE/FXRSTOR optimizations
pdpe1gb – GB pages
rdtscp – RDTSCP
lm – Long Mode (x86-64)
3dnowext – AMD 3DNow! extensions
3dnow – 3DNow!
k8 – Opteron, Athlon64
k7 – Athlon
constant_tsc – TSC ticks at a constant rate
up – smp kernel running on up
pebs – Precise-Event Based Sampling
bts – Branch Trace Store
nonstop_tsc – TSC does not stop in C states
pni – SSE-3
pclmulqdq – PCLMULQDQ instruction
dtes64 – 64-bit Debug Store
monitor – Monitor/Mwait support
ds_cpl – CPL Qual. Debug Store
vmx – Hardware virtualization
smx – Safer mode
est – Enhanced SpeedStep
tm2 – Thermal Monitor 2
ssse3 – Supplemental SSE-3
cid – Context ID
fma – Fused multiply-add
cx16 – CMPXCHG16B
xptr – Send Task Priority Messages
pdcm – Performance Capabilities
dca – Direct Cache Access
sse4_1 – SSE-4.1
sse4_2 – SSE-4.2
x2apic – x2APIC
aes – AES instructions
avx – Advanced Vector Extensions
hypervisor – Running on a hypervisor
lahf_lm – LAHF/SAHF in long mode
cmp_legacy – If yes HyperThreading not valid
svm – Secure virtual machine
extapic – Extended APIC space
cr8legacy – CR8 in 32-bit mode
abm – Advanced bit manipulation
sse4a – SSE-4A
ibs – Instruction Based Sampling
sse5 – SSE-5
skinit – SKINIT/STGI instructions
wdt – Watchdog timer

Happy Fish

Another Middle East undersea fiber cut

Internet service in Egypt, the UAE, India and other counties in the region was interrupted today when submarine cables in the Mediterranean Sea failed. 

Cause of the failure is not known at this time, says operator Reliance GlobalCom. A total of three cables linking Italy and Egypt went down. A ship has been sent out to fix the problems.

Currently, services are being rerouted to backup cables and satellites. As rerouting measures fully kick in, services are expected to improve by early Saturday morning.

This isn’t the first time that cable service to the region has failed. Earlier this year, five separate undersea cables were cut between late January and early February, interrupting service to a number of countries.  

There’s likely to be some serious failure analysis involved if we see some more pipes go off-line in the next week.

(quoted from here).

SANS Internet Storm Center page about this event.

Source NAT on Solaris 10 5/08 quick how-to

Enable ipv4 forwarding:

routeadm -u -e ipv4-forwarding

Configure NAT in /etc/ipf/ipnat.conf (replace e1000g0 with your outside interface, with your LAN IP range):

map e1000g0 -> 0/32 proxy port ftp ftp/tcp
map e1000g0 -> 0/32 portmap tcp/udp auto
map e1000g0 -> 0/32

Enable ipf:

svcadm enable network/ipfilter

Internet in Qatar

Starting this morning the Internet link is very slow, with 10-40% packet loss and 1.5 second RTT. This could be due to Qtel’s PCCWBTN link down starting after 8 a.m. AST. Traffic is routed via AS6453 (Teleglobe), which is heavily overloaded and oversubscribed. Of course, the FLAG Telecom link is not used for ADSL subscribers. So, the 2 MBPS ADSL link (largest residential package available in Qatar) real speed is around 200 kbps. This reminds me of home Internet access ten years ago.

2008 submarine cable disruption

Mediterranean Submarine Cables Cut

Two separate oceanic cable systems in the far East were severed around 0800 hrs GMT on January 30 2008, greatly impacting both Internet and voice traffic to the region. A third cable was cut at 0559 hrs GMT on February 1 2008.