Noel J. Bergman dijo: >> Hmm. The new machines are right for the work. AFAIK, one of them is >> intended to be used only as mail server. > > The new mail server is an IBM xSeries with dual Xeon processors and 2GB > RAM. > The old mail server is a dual Pentium III. Under the current "normal" > load, > which includes dealing with spam, I rarely see less than 10% usage. Often > it will spike to over 50% usage. I just watched the system get entirely > consumed for a few minutes. And that is on a Friday night with light > load, > and without a major worm
The percentages are related to the old or to the new server? If they are related to the old server we can divide by 4 the values, right? >> Instead of being afraid of the load, I think we really need to test it >> and >> see if the new machines can work with spamassassin or not. > > That's already on the docket. :-D >> > There is no reason in wasting precious CPU power for that. > >> precious CPU power? I really want to see where we use it. >> From minotaur, right now > > minotaur isn't the mail server. And average isn't te problem. I know minotaur is not running the main server. But it runs a MTA. We get some mail from it. I discover it because when daedalus was blocked by spamcop I was able to get mail to the Cocoon PMC maillist. I wonder why and discovered that mail was sent from minotaur. But what about distribute the load of spamassassin? AFAIK, with the "spamc" we can be able to use other machine to analyze the mail. I think it is posible. Of course if there is other machine that has a low load. > The problem is when the worms are running. I've got to chart it out > sometime, but we have been seeing a 4x higher than normal rate of spam for > a > while now. But that is a fraction of what we've seen when things get > really > bad, and the mail server is backed up for hours. I know what happen when things goes really, really bad. And I can tell you I saw often how well they manage the situation. I had experiences deploying anti-virus+anti-spam appliances before, on and after a peak virus and/or spam epidemias. In "nano" sized companies sometimes you see the one-man shows: he sell, drive, develop, deploy, configure, fix, teach, etc. ;-) > Personally, I am in favor of blocking all DHCP pools at the interface, and > not allowing direct connections. That is one of the lowest cost means of > stopping spambots running on Windows. If they want to send e-mail, they > should be told to go through their ISP, company server, or get a static > IP. Not sure. DHCP pools are still used for many people around the world and can make a "digital-divide world" (the a new fashion meme ;-). Seriously, I think lot of people is using fixed IP nows, but this will not be fine for a far more lot people that connect mainly via DHCP pools. For them the only solution is to share the IPs. About spambots on windows and other I really will appreciate to be able to view the current mailserver config in the ASF and see if I can help there. Please don't be ofended about this request. I know the ASF members managing the infraestructure is very competent, but I think we can apply the power of OSS development in the ASF server config. WDYT? Best Regards, Antonio Gallardo. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
