[clamav-users] We STILL cannot reliably get virus updates (since new mirrors)
h.reindl at thelounge.net
Mon Jul 2 13:50:55 EDT 2018
Am 02.07.2018 um 19:45 schrieb Brian Morrison:
> On Mon, 2 Jul 2018 19:17:32 +0200
> Reindl Harald wrote:
>> Am 02.07.2018 um 19:07 schrieb Brian Morrison:
>>> On Mon, 2 Jul 2018 10:26:34 +0200
>>> Reindl Harald wrote:
>>>> Am 02.07.2018 um 08:44 schrieb Bill Maidment:
>>>>> Maybe these are dumb questions; if so, please ignore.
>>>>> But doesn't it make more sense to update all the mirrors first,
>>>>> before changing the DNS? Is there some mechanism to do it that way
>>>> i wonder why all the linux distributions with update mirrors don't
>>>> need that DNS theatre to start with....
>>> Because the rate of updates is much less frequent, the more often
>>> you need to check the higher the mirror load becomes. Much of this
>>> load is telling people that there is no newer version...
>> says who?
> I am basing my comments on the distributed.net experience during the
> mid to late 90s. At the time they used one of the first DNS-based TXT
> record update mechanisms, it's broadly similar to how all of these work.
> At the time this made a very big difference to the load their mirrors
> were dealing with.
>> on a typical setup freshclam is running once or twice *daily* while a
>> webserver these days can spit out the same small static txt file many
>> thousands of times per seond with zero load
> For me freshclam runs roughly every 2 hours, so I think that the load
> is an order of magnitude higher than you state. I will confess that I
> don't know about the capability of web servers in this regard, but the
> point that d.net made was that the DNS server would always be more
> capable in this regard than a web server
come on - our main-server running ina virtual machine spits out 30000
requests/sec. on our core-cms in case of cache-hits and even on a 7
years old workstation far above 10000/sec and that is *not* static
content with a few bytes
More information about the clamav-users