Thanks for the explanation.
My original idea was that once we move the big mirrors (even a small
one would do) away to ISOC, we would have enough resources to host a
Clam DB on beak, where we do have administrative privileges.
I did not think about the bandwidth issue, but I believe it should be
less problematic than a distro mirror (note that we're talking about a
virusdef-db, not download mirror. Downloads of clam are handled by the
sourceforge mirror system)
-------
And now for some ideological babble (the kind we love to argue about
in hamakor lists ;-))
Why Clam is a good cause for the community to support:
* In the AV field, an open source alternative is highly important, as
it provides some counter to the inherent conflict of interests that
any commercial AV developer has.
* A propriatary alternative (you can guess which one) is distributed
pre-installed (at least on all new laptops I've seen recently), much
like a certain OS, and is actively (and annoyingly) pushed by (some?)
ISP's.
* Popularity of ClamWin in the "Wide World of Win...s users" might
increase awareness to availability and quality of open source software
in general.
Once the "on-access scanning component" (yuck...) is released, Clam
would become a serious competitor in the field, and faster db updates
would certainly give it an edge.
* (for those of you that oppose any software development for
propriatary OS'es) - Win...s viruses affect all internet users (less
infected Win...s machines mean less spammy infected emails hitting
your mailbox and eating up bandwidth).
Best,
Amit
On 6/7/07, Shachar Shemesh <[EMAIL PROTECTED]> wrote:
Amit Aronovitch wrote:
> On 6/6/07, Lior Kaplan <[EMAIL PROTECTED]> wrote:
>> Amit Aronovitch wrote:
>> > 1)
>> > Do you have download statistics?
>> > Personally I'm using the deb x86-64 mirror on (almost) daily basis,
>> > but I realize that we should address the needs of the community rather
>> > than those of the readers of this list...
>>
>> http://mirror.hamakor.org.il/#news
>
> I guess you refer to the move into ISOC servers.
> But out of curiousity - you probably have some historic download
> statistics for the mirrors we currently have - are they available
> online?
>
>>
>> > 2)
>> > In case we get over the problems and get some space for new mirrors -
>> > how about setting up a ClamAV database mirror?
>> > I understand that it has very low disk usage (50MB?), but it does
>> > require special settings (they use a push-mirroring system).
>> > http://www.clamav.net/doc/mirrors/clamav-mirror-howto.pdf
>>
>> I've read it, the require you to give them access to your machine for
>> "pushing" updates. I don't love that...
>
> I know. That's why I bothered mentioning it's a push system.
>
> However, just note that (non-shell) ssh access for an unprivileged
> special account (which would probably be chrooted too, assuming
> Shachar manages the server ;-) ) is not much of a security risk. I
> assume you already have ssh access for some (shell enabled) user
> accounts anyway.
>
> thanks,
> Amit
We need to understand a few things. On the ISOC server, we are very far
from free to install whatever service/user we want. Lior has shell
access, and that's about the extent of things allowed. Not only is a
"push" system out of the question, even running rsync for gentoo mirrors
is not possible.
The other thing we need to understand is that we CAN solve the server
issues. I'm not sure how much will it take to get the current server
fixed, but anywhere between 4K and 6K NIS is practically guaranteed to
get us a working server (this one or a new one), which will have about
800-900GB of free space. That's not where our trouble lies.
The real trouble is that mirror servers are bandwidth eaters, and as
such, are expensive to host. Actcom was willing to turn a blind eye
(except when we ate up 80% of its IIX connection due to three live CDs
being published in Yedioth Aharonot). It is obvious that Bezeq-intl will
not continue this policy, and it is doubtful whether we will find any
other ISP in Israel that will. Without friendly hosting, it is doubtful
whether Hamakor can afford the costs of a high bandwidth hosting! The
prices of regular hosting are about 100$/mo. We can afford that. I'm not
yet sure what the prices for high bandwidth hosting may be, and that's
an important question.
And that's the real appeal of the ISOC server. It resides in a strategic
location, and hosted by a partner that is strategic not only to us, but
also to the ISPs. It can get away with things we cannot.
Which means that the current planned action is that either beak or a
replacement server continue to host the relevant mailing lists, sites
and databases, but will no longer host mirrors. All mirrors will move to
ISOC or any other friendly hosting sponsor we may come up with.
Shachar
