Re: [clamav-users] How to pick / specify the correct mirror

2013-01-30 Thread Benny Pedersen

Noel Jones skrev den 28-01-2013 23:10:


Regardless whether you're using cron or daemonized, the default
behavior is to check DNS for the current version to decide if a
download is needed.


ups my fault, i got it now, thanks Noel



___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-29 Thread Jerry
On Mon, 28 Jan 2013 22:45:31 -0800
Dennis Peterson articulated:

 Bill Landry's unofficial sigs script does it right by using a random 
 function as an offset from cron time to actually initial a sig
 download. For those of us with a couple dozen or more systems this is
 a mirror-friendly way of not getting into lockstep with any
 fixed-offset scheme. Here is the code segment from Bill's script and
 I recommend something like it.
 
 if [ -n $RANDOM ]
then
   sleep_time=$(($RANDOM * $(($max_sleep_time -
 $min_sleep_time)) / 32767 + $min_sleep_time))
else
   sleep_time=0
   while [ $sleep_time -lt $min_sleep_time -o
 $sleep_time -gt $max_sleep_time ] ; do
  sleep_time=`head -1 /dev/urandom | cksum | awk '{print
 $2}'` done
 fi
 
 A bit of precision overkill given there are only 3600 seconds in an 
 hour, but it works. I use sleep $[ RANDOM % 900 ] and I have no idea
 nor care when the signature download actually begins, but every
 machine's crontab starts at the same time. The important thing to
 remember here is there is a finite amount of time required to dl all
 the sigs so regardless of the start time there will be overlap with
 other clients just because  there are more clients than there are
 seconds in an hour.

I use a much simpler function:

function get_files ()
{
# This is the snooze test. It will delay the start of a download session
# between 0 and 546 seconds (Roughly 9 minutes). It will only work when the
# script is run via CRON. The variable $REST activates this function.

# See if the variable 'REST' has been set or if forcing it from command line
# And if we are running via CRON
if [ ${REST} -gt 0 -a ! -t 0 ]; then
# Use the BASH RANDOM function to generate a random number between 0  32767
  RESTING=$((RANDOM/60))
  sleep ${RESTING}
fi

...
}

I never saw the benefit of expanding the time span, although it could
be done easily enough.

-- 
Jerry ♔

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the Reply-To header.
__

___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml

Re: [clamav-users] How to pick / specify the correct mirror

2013-01-28 Thread Benny Pedersen

Jim Preston skrev den 26-01-2013 03:39:

I am using freshclam but NOT running it as a daemon. Here is my cron 
task:

10 * * * * /usr/local/bin/freshclam /dev/null 21


this will be hardcoded time checks, not a fail if one can live with 
that



So as far as I understand, I am using freshclam. If I am wrong or
loose some functionality with the cron task, can you elaborate or
point me to the relevant documentation?


if you used freshclam as a deamon it will update when dns is showing 
new versions, not wait one hour or more in cron to get the newest 
updates




___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-28 Thread Noel Jones
On 1/28/2013 2:27 PM, Benny Pedersen wrote:

 if you used freshclam as a deamon it will update when dns is showing
 new versions, not wait one hour or more in cron to get the newest
 updates


Incorrect.

When you run freshclam as a daemon, the freshclam.conf specifies how
many times per day to check for updates.  There's not a lot of
practical difference between a daemonized freshclam with Checks 24
and a once-an-hour cron job.

The default is Checks 12 which means {check for an update 12 times
per day}.

Regardless whether you're using cron or daemonized, the default
behavior is to check DNS for the current version to decide if a
download is needed.


  -- Noel Jones
___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-28 Thread McDonald, Dan


On 1/28/13 4:10 PM, Noel Jones njo...@megan.vbhcs.org wrote:

 On 1/28/2013 2:27 PM, Benny Pedersen wrote:
 
 if you used freshclam as a deamon it will update when dns is showing
 new versions, not wait one hour or more in cron to get the newest
 updates
 
 
 Incorrect.
 
 When you run freshclam as a daemon, the freshclam.conf specifies how
 many times per day to check for updates.  There's not a lot of
 practical difference between a daemonized freshclam with Checks 24
 and a once-an-hour cron job.

I normally run checks 47 or checks 49.  That's not quite (or just
slightly more often than) every half hour.  Using an odd number like that
ensures that I don't get stuck on a popular minute for polls...

-- 
Daniel J McDonald, CCIE # 2495, CISSP # 78281

___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-28 Thread Dennis Peterson
Bill Landry's unofficial sigs script does it right by using a random 
function as an offset from cron time to actually initial a sig download. 
For those of us with a couple dozen or more systems this is a 
mirror-friendly way of not getting into lockstep with any fixed-offset 
scheme. Here is the code segment from Bill's script and I recommend 
something like it.


   if [ -n $RANDOM ]
  then
 sleep_time=$(($RANDOM * $(($max_sleep_time - $min_sleep_time)) 
/ 32767 + $min_sleep_time))

  else
 sleep_time=0
 while [ $sleep_time -lt $min_sleep_time -o $sleep_time 
-gt $max_sleep_time ] ; do

sleep_time=`head -1 /dev/urandom | cksum | awk '{print $2}'`
 done
   fi

A bit of precision overkill given there are only 3600 seconds in an 
hour, but it works. I use sleep $[ RANDOM % 900 ] and I have no idea nor 
care when the signature download actually begins, but every machine's 
crontab starts at the same time. The important thing to remember here is 
there is a finite amount of time required to dl all the sigs so 
regardless of the start time there will be overlap with other clients 
just because  there are more clients than there are seconds in an hour. 
On my personal servers I do this once each day and most of what I catch 
are traced to signatures from Sane Security. Less than 15/day, on average.


dp

On 1/28/13 2:50 PM, McDonald, Dan wrote:


On 1/28/13 4:10 PM, Noel Jones njo...@megan.vbhcs.org wrote:


On 1/28/2013 2:27 PM, Benny Pedersen wrote:


if you used freshclam as a deamon it will update when dns is showing
new versions, not wait one hour or more in cron to get the newest
updates


Incorrect.

When you run freshclam as a daemon, the freshclam.conf specifies how
many times per day to check for updates.  There's not a lot of
practical difference between a daemonized freshclam with Checks 24
and a once-an-hour cron job.

I normally run checks 47 or checks 49.  That's not quite (or just
slightly more often than) every half hour.  Using an odd number like that
ensures that I don't get stuck on a popular minute for polls...



___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Lee Graber
Thanks Jim. Is this different than db.local.clamav.net? That is what my
conf file was pre-populated with when I installed the package. I changed it
to us but I am wondering if that is actually changing anything. Thank for
the tip on cron task. That seems like a good idea. :)

On Thu, Jan 24, 2013 at 2:53 PM, Jim Preston jimli...@commspeed.net wrote:

 On 01/24/2013 11:40 AM, Lee Graber wrote:

 I am just starting to try and get clamd + freshclam running on some Ubuntu
 servers running on EC2 servers in the US Standard Region (east coast). The
 documentation talks about specifying a mirror which is close to you but it
 seems to default to the round-robin endpoint. I have run freshclam a
 number
 of times and sometimes it takes 14 seconds and sometimes it takes 6
 minutes. I can see that the ip address is different on the calls and am
 trying to figure out how I can get it to take 14 seconds all the time :).
 I
 can't find anything that says there is a mirror in the Amazon DataCenter.
 How can I make this behavior reliable fast.

 Thanks
 Lee
 __**_
 Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
 http://www.clamav.net/support/**ml http://www.clamav.net/support/ml

  You should have set up freshclam.conf to use the US mirrors
 (DatabaseMirror db.us.clamav.net). The update time will vary depending on
 whether it has to download any updates and the load on the particular
 mirror it connects to. The round-robin functionality is by design to spread
 load and help when a mirror goes down for any reason.

 You may also want to change the update frequency. I am running freshclam
 as a cron task and set it to not be on the hour, e.g. I update at 5 minutes
 after the hour to try and hit the mirrors at a low load time.


 --
 Jim Preston



 __**_
 Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
 http://www.clamav.net/support/**ml http://www.clamav.net/support/ml

___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Al Varnell
From my experience, it changes very little, especially in the US where there
are thirteen mirror sites.  I'm told that really isn't enough to meet the
demand here, so there are several off-shore sites added to the mix, which I
suppose could account for your more lengthy sessions.  Changing it to US
won't eliminate those off-shore sites from occasionally being selected in
the round-robin scheme of things.

The freshclam algorithm attempts to determine a set of sites closest to you
by examining your IP address.  Not a perfect solution, but it usually works
to be close enough.

In my work with ClamXav users, I normally try to discourage them from
changing the setting unless they are experiencing a high rate of
connectivity issues with sites that they probably should not be using.
Doing so is not a standard ClamXav preference option and changing the
freshclam.conf file is not a trivial task for the average Mac user.  There
have been occasions in Europe where a mirror site has prevented access from
certain IP blocks, which they aren't supposed to do, but there is really no
way to prevent it.  In that case, there did seem to be improvement by
changing the CC.

Of course, if you are going to be checking for updates more than once an
hour, then you are required to change it for load balancing purposes.
Again, this is not something the average Mac user can do (other than
manually), even if they wanted to, but enterprise sites must play by
different rules.

-Al-

On 1/25/13 12:54 AM, Lee Graber  wrote:

 Thanks Jim. Is this different than db.local.clamav.net? That is what my
 conf file was pre-populated with when I installed the package. I changed it
 to us but I am wondering if that is actually changing anything. Thank for
 the tip on cron task. That seems like a good idea. :)
 
 On Thu, Jan 24, 2013 at 2:53 PM, Jim Preston jimli...@commspeed.net wrote:
 
 On 01/24/2013 11:40 AM, Lee Graber wrote:
 
 I am just starting to try and get clamd + freshclam running on some Ubuntu
 servers running on EC2 servers in the US Standard Region (east coast). The
 documentation talks about specifying a mirror which is close to you but it
 seems to default to the round-robin endpoint. I have run freshclam a
 number
 of times and sometimes it takes 14 seconds and sometimes it takes 6
 minutes. I can see that the ip address is different on the calls and am
 trying to figure out how I can get it to take 14 seconds all the time :).
 I
 can't find anything that says there is a mirror in the Amazon DataCenter.
 How can I make this behavior reliable fast.
 
 Thanks
 Lee
 
 You should have set up freshclam.conf to use the US mirrors
 (DatabaseMirror db.us.clamav.net). The update time will vary depending on
 whether it has to download any updates and the load on the particular
 mirror it connects to. The round-robin functionality is by design to spread
 load and help when a mirror goes down for any reason.
 
 You may also want to change the update frequency. I am running freshclam
 as a cron task and set it to not be on the hour, e.g. I update at 5 minutes
 after the hour to try and hit the mirrors at a low load time.
 
 
 --
 Jim Preston




___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Benny Pedersen

Jim Preston skrev den 24-01-2013 23:53:


You may also want to change the update frequency. I am running
freshclam as a cron task and set it to not be on the hour, e.g. I
update at 5 minutes after the hour to try and hit the mirrors at a 
low

load time.


freshclam use dns to check if there is new updates, using cron you 
loose this functionality


so configure freshclam.conf to use 24 updates pr day, this will try to 
keep updates hourly, but if some mirror is down freshclam will not wait 
one hour to try another


show freshclam --list-mirrors if there is problems with some mirrors




___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Benny Pedersen

Lee Graber skrev den 25-01-2013 09:54:


That seems like a good idea. :)


nope understanding freshclam is even more worse :)

___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Lee Graber
Thanks Al,
   This is actually for a farm of servers (right now it is a farm of 1)
running Ubuntu on EC2. I was kinda surprised that there wasn't already a
mirror in AWS since they actually included ClamAV in their extended package
but such is life. My latest thoughts have been around having a single
server (or 2) in my farm whose job (among other things) is to run freshclam
and then push the updated cvd files (if needed) to a central S3 location.
Then all of the servers I have would simply pull from there when changes
are needed. This would mean less overhead on the mirrors and more
reliability on my servers all being able to update in a timely fashion. I
had thought about checking out the wire protocol for freshclam (I don't
think it is doc'd but I could use wireshark) so that I could maybe even
still run freshclam on my virus checking servers but I am a bit worried
that it could change and so I could just invent my own (it is basically
something along the lines of have any of these files changed. Give them to
me if they have. And then making sure clamd gets the changes). Has anyone
already built something like this? If I do this, I could put the cvds in a
public dataset on Amazon S3 so anyone inside AWS could get them for free
and fast but I am not sure about that yet. Again, is this something people
have done?

Thanks
Lee

On Fri, Jan 25, 2013 at 2:18 AM, Al Varnell alvarn...@mac.com wrote:

 shclam algorithm attempts to determine a set of sites closest to you
 by examining your IP address.  Not a perfect solution, but it usually works
 to be close enough.

 In my work with ClamXav users, I normally try to discourage them from
 changing the setting unless they are experiencing a high rate of
 connectivity issues with sites that they probably should not be using.
 Doing so is not a standard ClamXav preference option and changing the
 freshclam.conf file is not a trivial task for the average Mac user.  There
 have been occasions in Europe where a mirror site has prevented access from
 certain IP blocks, which they aren't supposed to do, but there is really no

___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Jim Preston

On 01/25/2013 10:03 AM, Benny Pedersen wrote:

Jim Preston skrev den 24-01-2013 23:53:


You may also want to change the update frequency. I am running
freshclam as a cron task and set it to not be on the hour, e.g. I
update at 5 minutes after the hour to try and hit the mirrors at a low
load time.


freshclam use dns to check if there is new updates, using cron you 
loose this functionality


so configure freshclam.conf to use 24 updates pr day, this will try to 
keep updates hourly, but if some mirror is down freshclam will not 
wait one hour to try another


show freshclam --list-mirrors if there is problems with some mirrors


Hi Benny,

I do not understand your comment freshclam use dns to check if there is 
new updates, using cron you loose this functionality


I am using freshclam but NOT running it as a daemon. Here is my cron task:
10 * * * * /usr/local/bin/freshclam /dev/null 21

So as far as I understand, I am using freshclam. If I am wrong or 
loose some functionality with the cron task, can you elaborate or point 
me to the relevant documentation?


Thanks, Jim

--
Jim Preston



___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Noel Jones
On 1/25/2013 8:39 PM, Jim Preston wrote:
 On 01/25/2013 10:03 AM, Benny Pedersen wrote:
 Jim Preston skrev den 24-01-2013 23:53:

 You may also want to change the update frequency. I am running
 freshclam as a cron task and set it to not be on the hour, e.g. I
 update at 5 minutes after the hour to try and hit the mirrors at
 a low
 load time.

 freshclam use dns to check if there is new updates, using cron you
 loose this functionality

 so configure freshclam.conf to use 24 updates pr day, this will
 try to keep updates hourly, but if some mirror is down freshclam
 will not wait one hour to try another

 show freshclam --list-mirrors if there is problems with some
 mirrors

 Hi Benny,
 
 I do not understand your comment freshclam use dns to check if
 there is new updates, using cron you loose this functionality
 
 I am using freshclam but NOT running it as a daemon. Here is my cron
 task:
 10 * * * * /usr/local/bin/freshclam /dev/null 21
 
 So as far as I understand, I am using freshclam. If I am wrong
 or loose some functionality with the cron task, can you elaborate or
 point me to the relevant documentation?
 
 Thanks, Jim
 


As long as you don't specify the --no-dns option, freshclam will use
DNS to compare the current published version before attempting to
download anything, regardless whether it's run by hand, a cronjob,
or daemonized.

At some point in the distant past, the freshclam daemon had a
tendency to misbehave/hang/crash/whatever, prompting some folks to
run it as a cron job.  This has been long fixed, and the background
daemon is the recommended method now, as it somewhat randomizes the
checking time to spread load on the download servers.  But you don't
lose anything by running it under cron.



  -- Noel Jones
___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-25 Thread Dennis Peterson

On 1/24/13 10:40 AM, Lee Graber wrote:

I am just starting to try and get clamd + freshclam running on some Ubuntu
servers running on EC2 servers in the US Standard Region (east coast). The
documentation talks about specifying a mirror which is close to you but it
seems to default to the round-robin endpoint. I have run freshclam a number
of times and sometimes it takes 14 seconds and sometimes it takes 6
minutes. I can see that the ip address is different on the calls and am
trying to figure out how I can get it to take 14 seconds all the time :). I
can't find anything that says there is a mirror in the Amazon DataCenter.
How can I make this behavior reliable fast.

Thanks
Lee
___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml
Since you are running a Unix system you can use your own BIND 
installation to create a freshclam zone of your choosing. Create a zone, 
say freshclam.grabner, populate it with IP addresses from existing 
reliable servers using my.freshclam.grabner in a round-robin array, and 
use that (my.freshclam.grabner) in your freshclam.conf file.


Obviously it becomes your responsibility to maintain an accurate list, 
but that can be entirely automated with some simple scripting. This 
isn't anything I would recommend, but there are gaps in the global 
coverage it can help fill, and it will certainly work while putting all 
the joy and heartache of success or failure in your camp.


dp


___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


[clamav-users] How to pick / specify the correct mirror

2013-01-24 Thread Lee Graber
I am just starting to try and get clamd + freshclam running on some Ubuntu
servers running on EC2 servers in the US Standard Region (east coast). The
documentation talks about specifying a mirror which is close to you but it
seems to default to the round-robin endpoint. I have run freshclam a number
of times and sometimes it takes 14 seconds and sometimes it takes 6
minutes. I can see that the ip address is different on the calls and am
trying to figure out how I can get it to take 14 seconds all the time :). I
can't find anything that says there is a mirror in the Amazon DataCenter.
How can I make this behavior reliable fast.

Thanks
Lee
___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml


Re: [clamav-users] How to pick / specify the correct mirror

2013-01-24 Thread Jim Preston

On 01/24/2013 11:40 AM, Lee Graber wrote:

I am just starting to try and get clamd + freshclam running on some Ubuntu
servers running on EC2 servers in the US Standard Region (east coast). The
documentation talks about specifying a mirror which is close to you but it
seems to default to the round-robin endpoint. I have run freshclam a number
of times and sometimes it takes 14 seconds and sometimes it takes 6
minutes. I can see that the ip address is different on the calls and am
trying to figure out how I can get it to take 14 seconds all the time :). I
can't find anything that says there is a mirror in the Amazon DataCenter.
How can I make this behavior reliable fast.

Thanks
Lee
___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml

You should have set up freshclam.conf to use the US mirrors 
(DatabaseMirror db.us.clamav.net). The update time will vary depending 
on whether it has to download any updates and the load on the particular 
mirror it connects to. The round-robin functionality is by design to 
spread load and help when a mirror goes down for any reason.


You may also want to change the update frequency. I am running freshclam 
as a cron task and set it to not be on the hour, e.g. I update at 5 
minutes after the hour to try and hit the mirrors at a low load time.



--
Jim Preston



___
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://www.clamav.net/support/ml