Comcast's caching resolvers appear to ignore TTLs. I've ran into this
on a couple occasions. :\
On Sat, 2004-08-21 at 17:24, Robert Blayzor wrote:
> maybe we should just have a ton of public mirrors
We already have a ton of public mirrors. ;)
Seriously though, if you'd like to host a mirror, c
Ok, so we have freshclam, but what about making the databases available
via rsync? rsync is a lot more efficient when transfering files like
this where just the "diffs" to the files are sent.
The problem with the DNS is how often people look for changes. If the
TTL is low and people monitor i
At 08:06 21.08.2004 -0600, you wrote:
That"s why the new method of using DNS
to cache the version number will be a much better solution and I think
it will allow more frequent checks without burdening the mirrors and
shifting the load to the DNS infrastructure.
I am not a DNS expert by any means an
At 08:06 21.08.2004 -0600, you wrote:
That"s why the new method of using DNS
to cache the version number will be a much better solution and I think
it will allow more frequent checks without burdening the mirrors and
shifting the load to the DNS infrastructure.
I am not a DNS expert by any means an
That"s why the new method of using DNS
to cache the version number will be a much better solution and I think
it will allow more frequent checks without burdening the mirrors and
shifting the load to the DNS infrastructure.
I am not a DNS expert by any means and I may be way off in left
field here
Tomasz Kojm wrote:
On Fri, 13 Aug 2004 22:04:31 +0200
Erich Titl <[EMAIL PROTECTED]> wrote:
become an issue). I believe if such a process can verify the md5sum of
the input file against a publicly known md5sum of the original input,
then to some extent a cdv file could be created locally. The si
Hi
At 21:41 13.08.2004, you wrote:
Erich Titl wanted us to know:
>Nice, we could actually build a wrapper around freshclam to only fetch when
>there is a new version ready.
It already does this.
Yes, but it uses TCP, not hierarchically distributed servers, all this has
been discussed lately. DNS i
Am Friday 13 August 2004 22:05 schrieb Damian Menscher:
Hi,
> other ideas. But still, checking DNS for an update means a single UDP
> packet each way (which might even get cached).
In the propose use case the DNS info is most probably cached by the next ISP
already.
> Downloading the first
On Fri, 13 Aug 2004 22:04:31 +0200
Erich Titl <[EMAIL PROTECTED]> wrote:
> become an issue). I believe if such a process can verify the md5sum of
> the input file against a publicly known md5sum of the original input,
> then to some extent a cdv file could be created locally. The signature
You do
Damian Menscher wanted us to know:
>Good to know, and maybe that means we're wasting our time on all these
>other ideas. But still, checking DNS for an update means a single UDP
>packet each way (which might even get cached).
Yeah, I can see the simplicity and advantage of such a method.
--
Re
On Fri, 13 Aug 2004, Todd Lyons wrote:
> Erich Titl wanted us to know:
>
> >Nice, we could actually build a wrapper around freshclam to only fetch when
> >there is a new version ready.
>
> It already does this.
>
> It only retrives the first 512 bytes of data from each CVD file. Here
> is what a s
At 21:13 13.08.2004, you wrote:
On Fri, 13 Aug 2004 at 19:48:34 +0200, Erich Titl wrote:
[...]
> 7) rebuild a .CVD file
>
> I have not been able to rebuild the cvd file using sigtool, so this is for
> someone with more sigtool experience, but the diff of the two files show
> that a patch is easily
Erich Titl wanted us to know:
>Nice, we could actually build a wrapper around freshclam to only fetch when
>there is a new version ready.
It already does this.
12:39:51.553344 IP 10.1.1.240.41996 > 65.77.42.207.80: P 1:145(144) ack 1 win 5840
0x: 4500 00c4 1b41 4000 4006 a6e6 0a
On Fri, 13 Aug 2004 at 19:48:34 +0200, Erich Titl wrote:
[...]
> 7) rebuild a .CVD file
>
> I have not been able to rebuild the cvd file using sigtool, so this is for
> someone with more sigtool experience, but the diff of the two files show
> that a patch is easily feasible
You can't rebuild a
At 13:17 13.08.2004, you wrote:
Hi,
Following the long thread of "Idea for more timely virusdb updates", I
have put together
a basic system of putting daily.cvd in DNS TXT records.
It stores current version of daily.cvd, new signatures, and what time a
particular signature was added.
Which means
On Fri, 13 Aug 2004 14:01:49 +0200
Tomasz Papszun <[EMAIL PROTECTED]> wrote:
On Fri, 13 Aug 2004 at 18:17:19 +0700, Fajar A. Nugraha
wrote:
I have put together
a basic system of putting daily.cvd in DNS TXT records.
[...]
Though lacking secure digital signing, quite interesting
real solution
of
On Fri, 13 Aug 2004 at 18:17:19 +0700, Fajar A. Nugraha wrote:
>
> Following the long thread of "Idea for more timely virusdb updates", I
> have put together
> a basic system of putting daily.cvd in DNS TXT records.
[...]
Though lacking secure digital signing, quite interesting real solution
of
17 matches
Mail list logo
