words and letters

June 21, 2018

~# nvidia-docker run -it -p 8888:8888 tensorflow/tensorflow:1.1.0-devel-gpu-py3
1.1.0-devel-gpu-py3: Pulling from tensorflow/tensorflow
c62795f78da9: Pull complete
d4fceeeb758e: Pull complete
5c9125a401ae: Pull complete
0062f774e994: Pull complete
6b33fd031fac: Pull complete
7ab5dd833cf2: Pull complete
df9cc763fcde: Pull complete
9b0174a3640e: Pull complete
1efd10acdd72: Pull complete
f77b671e3092: Pull complete
af9093817c44: Pull complete
b2fb381211f0: Pull complete
af9e216e9ceb: Pull complete
09860f91cae6: Pull complete
9f0f8d45045d: Pull complete
0467b7b5b3c8: Pull complete
311d6a376dba: Pull complete
d775db884640: Pull complete
befeab6b5e5d: Pull complete
7dd802ca32d9: Pull complete
d1a48577f024: Pull complete
9e3601307c1a: Pull complete
057615e69ff0: Pull complete
Digest: sha256:8eaa148ede85f6689aa3ee5c6d6d34e5b05fdc9af7c293a56f59887aedb851d2
Status: Downloaded newer image for tensorflow/tensorflow:1.1.0-devel-gpu-py3
nvidia-docker | 2017/10/17 10:32:09 Error: unsupported CUDA version: driver 6.5 < image 8.0.61

June 21, 2018

docker

when i do

~$ docker build –pull -t $USER/tensorflow-serving-devel -f Dockerfile.devel .

it doesn’t finish…stole Dockerfile.devel fromhttps://github.com/tensorflow/serving/blob/master/tensorflow_serving/tools/docker/Dockerfile.devel



FROM ubuntu:16.04

MAINTAINER Jeremiah Harmsen <[email protected]>

RUN apt-get update && apt-get install -y \
        build-essential \
        curl \
        git \
        libfreetype6-dev \
        libpng12-dev \
        libzmq3-dev \
        mlocate \
        pkg-config \
        python-dev \
        python-numpy \
        python-pip \
        software-properties-common \
        swig \
        zip \
        zlib1g-dev \
        libcurl3-dev \
        openjdk-8-jdk\
        openjdk-8-jre-headless \
        wget \
        && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

# Set up grpc

RUN pip install mock grpcio

# Set up Bazel.

ENV BAZELRC /root/.bazelrc
# Install the most recent bazel release.
ENV BAZEL_VERSION 0.5.1
WORKDIR /
RUN mkdir /bazel && \
    cd /bazel && \
    curl -fSsL -O https://github.com/bazelbuild/bazel/releases/download/$BAZEL_VERSION/bazel-$BAZEL_VERSION-installer-linux-x86_64.sh && \
    curl -fSsL -o /bazel/LICENSE.txt https://raw.githubusercontent.com/bazelbuild/bazel/master/LICENSE && \
    chmod +x bazel-*.sh && \
    ./bazel-$BAZEL_VERSION-installer-linux-x86_64.sh && \
    cd / && \
    rm -f /bazel/bazel-$BAZEL_VERSION-installer-linux-x86_64.sh

CMD ["/bin/bash"]

</[email protected]>

but its pulled…

# docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
nvidia/cuda latest 9d3688f974a1 4 days ago 1.73GB
hello-world latest 05a3bd381fc2 4 weeks ago 1.84kB
tensorflow/tensorflow latest-gpu 4ca1d30fcd9b 2 months ago 3.28GB
nvidia/cuda 6.5 794ce05b823b 10 months ago 1.02GB

because i using cuda 6.5 on my “host” the latest one doesnt work, have to get the one tagged 6.5

# nvidia-docker run –rm -ti nvidia/cuda:6.5 nvcc –version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2014 NVIDIA Corporation
Built on Wed_Aug_27_10:36:36_CDT_2014
Cuda compilation tools, release 6.5, V6.5.16

so what next?

June 21, 2018

cudnn90

shucks this is the max cuda for your nvidia driver version..
at least 361.00 is required for CUDA 8.0
at least 352.00 is required for CUDA 7.5
at least 346.00 is required for CUDA 7.0
at least 340.00 is required for CUDA 6.5
at least 331.00 is required for CUDA 6.0
at least 319.00 is required for CUDA 5.5

apt-get install ansible

and then

maybe save the below as cuda.yml and run ansible-playbook cuda.yml ?


- hosts: localhost
  #user: somebody
  #sudo: yes
  tasks:
    - name: register CUDA repository
      shell: wget https://developer.nvidia.com/compute/cuda/9.0/Prod/local_installers/cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64-deb && dpkg -i local_installers/cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64.deb

    - name: install CUDA
      apt: name=cuda update_cache=yes

    - name: write environmant variables to bash profile
      copy: src=./cuda.sh dest=/etc/profile.d/

    - pip:
        name: numpy

    - pip:
        name: Keras

    - pip:
        name: tensorflow-gpu

    #- pip:
    #    name: --upgrade $TF_BINARY_URL
    #  environment:
    #    CUDA_HOME: /cluster/apps/x86_64/packages/cuda_8.0
    #    TF_BINARY_URL: https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow-0.11.0rc1-cp27-none-linux_x86_64.whl

install cudnn

cudnn-9.0-linux-x64-v7

libcudnn7_7.0.2.43-1+cuda9.0_amd64

libcudnn7-dev_7.0.2.43-1+cuda9.0_amd64

cudnn-6.5-linux-x64-v2

cudnn-sample-v2

June 21, 2018

postfix

we have a problem in that we backscatter. so we get on blacklist.
so a few things we can do to reduce the issue.

we are more strict with what mails get sent..

in main.cf

reject_invalid_hostname,
reject_non_fqdn_hostname,
reject_non_fqdn_sender,
reject_non_fqdn_recipient,
reject_unlisted_recipient,
reject_unknown_sender_domain,
reject_unknown_recipient_domain,

we don’t send email if the client is on a rbl..

reject_rbl_client zen.spamhaus.org,
reject_rbl_client bl.spamcop.net,
reject_rbl_client b.barracudacentral.org,

#reject_rbl_client dnsbl.sorbs.net

#i don’t like sorbs, think too many false positives
smtpd_client_restrictions = reject_unknown_client_hostname, sleep 3, reject_unauth_pipelining
smtpd_sender_restrictions = reject_unknown_sender_domain

Reject_unknown_client_hostname is the most effective, but also the most stringent. It blocks connections from any machine that does not have proper DNS resolution. Nearly every legitimate mail server has correct forward and reverse resolution configured these days, so the likelihood of false positives is still pretty low.

The “sleep3, reject_unauth_pipelining” pair tell the server to wait three seconds after the client connects before displaying its “banner” in the SMTP exchange. Though their numbers are declining, some spam programs don’t wait for the banner and start spewing right away. “Reject_unauth_pipelining” tells Postfix to hang up the phone of any such clients.

“Reject_unknown_sender_domain” applies to the email address sent with the “MAIL FROM” command in the SMTP dialogue (known as the “envelope sender”). That rule requires the email address be @ a valid domain. Non-deliverable notices are sent to that address so it could have some, though I suspect small, effect of reducing NDNs to bogus addressees. Note that the sending server isn’t asked to validate that the message comes from a workign address (since it often can’t); the only check is whether the sender’s domain exists.

You should also browse the possibilities in: http://www.postfix.org/ADDRESS_VERIFICATION_README.html

we give up trying to resend bounces sooner?

and shorten bounce queue to 1m

and

we try to not respond to dsn?

in master.cf
-o smtpd_discard_ehlo_keywords=silent-discard,dsn

or in this case allow local subnet only…

/etc/postfix/main.cf:
smtpd_discard_ehlo_keyword_address_maps =
cidr:/etc/postfix/esmtp_access

/etc/postfix/esmtp_access:

# Allow DSN requests from local subnet only
172.20.0.0/16      silent-discard
0.0.0.0/0           silent-discard, dsn
::/0                silent-discard, dsn

we reject DSNs outright..this is still commented off…

In my Postfix configuration, on the machine that acts as MX for my domains — edit ‘/etc/postfix/header_checks’, and add these lines:

/^Content-Type: multipart\/report; report-type=delivery-status\;/ REJECT no third-party DSNs
/^Content-Type: message\/delivery-status; / REJECT no third-party DSNs

Edit ‘/etc/postfix/main.cf’, and ensure it contains:

header_checks = regexp:/etc/postfix/header_checks

Then run:`

sudo /etc/init.d/postfix restart

June 21, 2018

openafs-part2

iscsiadm -m discovery -t sendtargets -p 172.20.193.40 && sleep 15 && iscsiadm -m node -T iqn.2001-05.com.equallogic:4-42a846-4ee8a9f13-1480000025e58afe-afs50 –login && sleep 15 && mount -a
mailsend.pl

#!/usr/bin/perl -w
#
# Program: SMTP mail delivery program
# Author: [email protected]
# Date: 17th nov 2008
#
# This script takes an smtp server as its only argument, reads an email
# on standard input, and delivers it to the smtp server for relaying to
# the final destination. For example,
#
# % mailsend.pl smtp.mycompany.com < myemail.txt
#
# Where “myemail.txt” contains the necessary mail headers, as well as
# the body text. For example,
#
# From: [email protected]
# To: [email protected]
# Subject: hello world
# Date: Sun, 16 Nov 2008 16:25:38 +0800
#
# this is my message
#
# Modified by jeffrey 171209
# now can have multiple recipients in the To/Cc field
use Socket ;
# —————————————————————————–
my $timeout = 30 ; # smtp inactivity timeout (seconds)
my $line = “” ; # general purpose line buffer
my $host = “” ; # the remote host that we’ll connect to
my $mail_from = “” ; # “From:” pulled off mail headers
my $rcpt_to = “” ; # “To:” pulled off mail headers
my @header = () ; # array of lines, which form the mail headers
[email protected]_to;
# —————————————————————————–
# This function is supplied a email address in the format

get-mptstatus.sh

#!/bin/bash
# This script uses mpt-status to probe /dev/mptctl to find out status of the LSI HW raid array
MPTSTATUS=”/usr/local/bin/mpt-status”
SMTP_SERVER=”smtp.bii-sg.org”
if [ ! -x $MPTSTATUS ]; then
echo “cannot find $MPTSTATUS”
exit
fi
host $SMTP_SERVER | grep “has address”
if [ $? -ne 0 ]; then
SMTP_SERVER=”123.136.68.157″
fi
cat /proc/kallsyms | grep mptctl > /dev/null
# we check the kernel table symbols, not found then try to load as kernel module.
# so there is no need to modprobe if mptctl is compiled in 🙂
if [ $? -ne 0 ]; then
/sbin/modprobe mptctl
if [ $? -ne 0 ]; then
echo “mptctl module not found, please recompile kernel module”
echo “Look at – <kernelsource>/drivers/message/fusion/Kconfig for details”
exit
fi
fi
sleep 2
if [ ! -e /dev/mptctl ]; then
/bin/mknod /dev/mptctl c 10 220
fi
# get the right “-i <controller_id>”
RUNMPT=`$MPTSTATUS -p | grep ‘mpt-status -i’ | sed ‘s/^.*\(mpt-status -i [0-9]\).*$/\1/’`
STATE=`/usr/local/bin/$RUNMPT | grep -w OPTIMAL | sed ‘s/^.*\(OPTIMAL\).*$/\1/’`
if [ “$STATE” != “OPTIMAL” ]; then
echo “hardware raid degraded on `date`” >> /var/log/messages
# we call sendmail.pl here
HOSTNAME=`hostname -f`
echo “From: [email protected]$HOSTNAME” > /root/bin/alertmail.txt
echo “To: [email protected]” >> /root/bin/alertmail.txt
echo “Subject: mpt-status alert – x4100 hardware raid is degraded” >> /root/bin/alertmail.txt
echo ” ” >> /root/bin/alertmail.txt
echo “$HOSTNAME hardware raid in degraded mode – please replace HDD immediately” >> /root/bin/alertmail.txt
/root/bin/mailsend.pl $SMTP_SERVER < /root/bin/alertmail.txt
fi

iscsi.sh

#!/bin/bash
mount | grep sdc1 || /root/bin/mailsend.pl 123.136.68.157 < /root/bin/iscsinotmounted.txt
sleep 30 && echo $(date) > /vicepa/date || /root/bin/mailsend.pl 123.136.68.157 < /root/bin/cannotwrite2iscsi.txt
sleep 60 && rm -f /vicepa/date || /root/bin/mailsend.pl 123.136.68.157 < /root/bin/cannotwrite2iscsi.txt
# Joe Smith <[email protected]>
# and it returns :
# [email protected]
sub f_email_addr
{
my $addr = $_[0] ;
$addr =~ s/^.+\<//g ;
$addr =~ s/\>.*$//g ;
return ($addr) ;
}
# This function sends a line to the remote smtp server, and reads a reply.
# The first argument is expected to be an established tcp connection. The
# line is supplied as the second argument, and the expected smtp reply
# code (number) is specified as the third argument. If the supplied smtp
# code is received, we return an empty string, otherwise something is
# presumed to have gone wrong and we return the entire reply string.
# Finally, an inactivity timeout is enforced. If triggered, the string
# “inactivity timed out” is returned.
sub f_smtp_dialog
{
my $fd = $_[0] ;
my $msg = $_[1] ;
my $expect = $_[2] ;
my $buf = “” ;
# deliver a message to the remote smtp server, if one is specified.
if (length ($msg) > 0)
{ printf ($fd “$msg”) ; }
eval
{
alarm ($timeout) ;
$buf = <$fd> ;
alarm (0) ;
} ;
if ($buf =~ /^$expect/)
{ return (“”) ; }
chomp ($buf) ;
return ($buf) ;
}
# —————————————————————————–
# check usage first …
if ($#ARGV != 0)
{
printf (“Usage : %s <smtp server> < mail.txt\n”, $0) ;
exit (1) ;
}
# try to read in mail header from stdin.
while ($line = <STDIN>)
{
chomp ($line) ;
if ($line =~ /^To:/)
{
$rcpt_to = $line ;
$rcpt_to =~ s/To: // ;
[email protected]=split(/,/,$rcpt_to);
foreach(@recp){
push(@recp_to, f_email_addr($_));
}
}
if ($line =~ /^Cc:/)
{
my$tline;
$tline=$line ;
$tline=~ s/Cc: // ;
[email protected]=split(/,/,$tline);
foreach(@recp){
push(@recp_to, f_email_addr($_));
}
}
if ($line =~ /^From:/)
{
$mail_from = $line ;
$mail_from =~ s/From: // ;
$mail_from = f_email_addr ($mail_from) ;
}
if ($line =~ /^$/)
{
last ;
}
$header[$#header+1] = $line ;
}
# check that we know who we’re sending mail from, and to.
if (length ($mail_from) < 1)
{
printf (“FATAL! Do not know who this mail is from. No ‘From: xx’.\n”) ;
exit (1) ;
}
if ($#recp_to < 0)
{
printf (“FATAL! Do not know who this mail is to. No ‘To: xx’.\n”) ;
exit (1) ;
}
# now attempt to connect to smtp server.
$host = inet_aton ($ARGV[0]) ||
die (“FATAL! Host not found $ARGV[0],$!”) ;
socket ($SD, PF_INET, SOCK_STREAM, getprotobyname (“tcp”)) ||
die (“FATAL! socket() failed,$!”) ;
eval
{
local $SIG{ALRM} = sub
{ printf (“FATAL! Connect to $ARGV[0] timed out.\n”) ; exit (1) ; } ;
alarm ($timeout) ;
connect ($SD, sockaddr_in (25, inet_aton ($ARGV[0]))) ||
die (“FATAL! Cannot connect to $ARGV[0] on port 25,$!”) ;
alarm (0) ;
} ;
select ($SD) ;
$| = 1 ;
select (STDOUT) ;
# perform the standard dialog with the smtp server.
$line = f_smtp_dialog ($SD, “”, 220) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; exit (1) ; }
$line = f_smtp_dialog ($SD, “HELO localhost\n”, 250) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; exit (1) ; }
$line = f_smtp_dialog ($SD, “MAIL FROM: $mail_from\n”, 250) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; exit (1) ; }
#$line = f_smtp_dialog ($SD, “RCPT TO: $rcpt_to\n”, 250) ;
#if (length ($line) > 0)
# { printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; exit (1) ; }
foreach(@recp_to){
my$k=$_;
#print “k $k\n”;
$line = f_smtp_dialog ($SD, “RCPT TO: $k\n”, 250) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; exit (1) ; }
}
$line = f_smtp_dialog ($SD, “DATA\n”, 354) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; exit (1) ; }
# now print out the mail header and read the rest of the mail body.
foreach (@header)
{ print ($SD “$_\n”) ; }
print ($SD “\n”) ;
while ($line = <STDIN>)
{ print ($SD “$line”) ; }
$line = f_smtp_dialog ($SD, “.\n”, 250) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; }
$line = f_smtp_dialog ($SD, “QUIT\n”, 221) ;
if (length ($line) > 0)
{ printf (“FATAL! $ARGV[0] said: %s\n”, $line) ; }
exit (0) ;

June 21, 2018

works for osx 10.5 ppc g4 7450

macbrowsers/WebKit-602.4.8_3-Leopard-PowerPC.dmg

macbrowsers/TenFourFox7450-45.8.0.app.zip

macbrowsers/seamonkey-for-ppc-2.26-RELEASE.app.zip

June 21, 2018

openafs-part1

yum groupinstall -y “Development Tools”, on my vagrant box scalefactory/centos6 , need to isntall ncurses too.
yum update -y
download openafs server from openafs website OpenAFS 1.6.20.1, 16MB
./configure –prefix=/usr/local/packages/openafs
make, make install
copy /etc/openafs from existing server
in /etc/openafs/server, there is

1

2

3

4

5

6

7

i. CellServDB

>example.com

10.0.2.15

ii. ThisCell

srv.example.com

iii.UserList

afsadmin
in /etc/openafs there is cacheinfo

1

/afs:/usr/local/packages/cache:500000

might want to make a folder called /usr/local/packages/cache for it..

its dangerous to go alone, take this:

#!/bin/bash
#
# Graceful AFS start and stop script.
# Written by: [email protected]
# Date 5th July 2008
STOP_RETRIES=10 # seconds to attempt afs shutdown
AFS=/usr/local/packages/openafs # where afs lives – symblink to current afs ver
AFS_DB=”afs60 afs61 afs62″ # list of afs database servers
case $1 in
‘start’)
# Get a list of afs database servers first.
DB=””
REACHABLE=0
CELLSERVDB=”$AFS/etc/openafs/CellServDB”
for I in $AFS_DB ; do
HOST=`grep $I $CELLSERVDB | awk ‘{print $1}’`
if [ -z “$HOST” ] ; then
echo “FATAL! Cannot locate $I in $CELLSERVDB”
exit 1
fi
# do a simple ping test first.
if [ “`uname -s`” = “SunOS” ] ; then
ping -n -s $HOST 40 3 >/dev/null 2>/dev/null
else
ping -n -c 3 $HOST >/dev/null 2>/dev/null
fi
if [ $? -eq 0 ] ; then
echo “$HOST ping test passed.”
# use “bos” to test if afs database services are running.
BOS=$AFS/bin/bos
N=`$BOS status $HOST -noauth | grep running | wc -l | awk ‘{print $1}’`
if [ $N -ge 4 ] ; then
echo “$HOST bos status passed.”
REACHABLE=1
break
else
echo “WARNING! AFS services not running normally on $HOST.”
fi
else
echo “WARNING! $HOST is not responding to ping.”
fi
done
# only fire up AFS if at least 1 database server is reachable.
if [ $REACHABLE -eq 0 ] ; then
echo “FATAL! No AFS database servers reachable.”
exit 1
fi
# check afs system call number (Solaris only) and load kernel module.
if [ “`uname -s`” = “SunOS” ] ; then
SYSNUM=`grep afs /etc/name_to_sysnum | awk ‘{print $2}’`
if [ -z “$SYSNUM” ] ; then
echo “FATAL! AFS name_to_sysnum not configured.”
exit 1
else
if [ $SYSNUM -ne 65 ] ; then
echo “FATAL! AFS name_to_sysnum not 65.”
exit 1
fi
fi
modload /kernel/fs/afs
if [ $? -ne 0 ] ; then
echo “FATAL! Failed to load AFS kernel module.”
exit 1
fi
fi
# on Linux we do a insmod
if [ “`uname -s`” = “Linux” ] ; then
insmod $AFS/lib/openafs/libafs.ko
if [ $? -ne 0 ] ; then
echo “FATAL! Failed to load AFS kernel module.”
exit 1
fi
fi
# the very last check is to make sure that /afs exists.
if [ ! -d /afs ] ; then
mkdir /afs
if [ $? -ne 0 ] ; then
exit 1
fi
fi
# if we have a dedicated afs cache filesystem, mount it now,
if [ -b /dev/zvol/dsk/mud/afs-cache ] ; then
if [ ! -d /var/afs-cache ] ; then
mkdir /var/afs-cache
chmod 700 /var/afs-cache
fi
mount -o nologging /dev/zvol/dsk/mud/afs-cache /var/afs-cache
fi
# finally, check whether to use memory cache or not.
mount | grep ‘/var/afs-cache’ >/dev/null
if [ $? -eq 0 ] ; then
$AFS/sbin/afsd
else
echo “starting with memcache”
$AFS/sbin/afsd -memcache
fi
# allow setuid bit on /afs.
if [ $? -eq 0 ] ; then
$AFS/bin/fs setcell bii.a-star.edu.sg -suid
fi
;;
‘stop’)
# check if /afs is mounted, if not mounted, don’t bother.
mount | grep AFS >/dev/null
if [ “$?” != “0” ] ; then
“FATAL! /afs is not mounted.”
exit 1
fi
# check for processes using /afs, kill them before we continue.
RETRIES=$STOP_RETRIES
HALF_WAY=$(($STOP_RETRIES / 2))
while [ $RETRIES -gt 0 ] ; do
PIDS=`fuser /afs 2>/dev/null | cut -f 2- -d ‘ ‘ | sed -e ‘s/[^0-9 \t]//g’`
if [ -z “$PIDS” ] ; then
break
fi
if [ $RETRIES -gt $HALF_WAY ] ; then
kill $PIDS
else
kill -9 $PIDS
fi
sleep 1
RETRIES=$(($RETRIES – 1))
done
# at this point, it *should* be safe to umount /afs.
$AFS/bin/fs flush
umount /afs
/sbin/rmmod libafs
;;
‘check’)
$AFS/bin/fs checkservers
$AFS/bin/fs checkvolumes
;;
*)
echo “Usage : $0 { start | stop | check }”
;;
esac

June 21, 2018

Hello world!

Welcome to WordPress. This is your first post. Edit or delete it, then start writing!

Posts

procmailrc

how is amber related?

docker2

docker

cudnn90

postfix

openafs-part2

works for osx 10.5 ppc g4 7450

openafs-part1

Hello world!