Using this to jot down some notes, that I will add to my darkmilieu.org website. This is NOT intended to be a 'MS Sucks' theme, but would like to air some of my problems, and see if others have experienced the same....
Project: Implement Postfix MX server for MSEx Customer
Problem: Customer has POP3 users, and an open relay.
Solution: As the platform is Ex2k, I can't go the POP User route as with 5.5. Support.microsoft.com (smc) gives the "solution" of starting an SMTP instance on a high port for POP users. I find this a bit amateurish, and figure to do a Postfix Mail Exchanger. This may be extensive, however the customer wishes to have mail stored should the Ex2k server go down (customer has some experience w/ ex2k ;-) Postfix will handle this. So here is a BRIEF summary of the "issues" encountered.
o RH8.0 has been installed. Postfix installed, configured, and prestaged.
o Maint Window is open, start finishing touches, everything goes smoothly. pam_smb works great, and the DC isn't giving us problems.
o Test SMTP works
o move to testing with relay restrictions works ... skip ahead...
o all testing done. changine dns, and moving to production. Smooth.
o Test... POP3 is hanging.
o restart restart POP3 FAILS Error 1053
o search smc.. nada for POP.
o options to do anything in services for POP are
o ExcSysmgr has stop option avail.
o Stop service.. FAILS, no error nothing in eventvwr. Message indicates that Service is stopped, that it must be started.
o Start option is greyed... reboot.
o System comes up (meanwhile postfix is storing the mail wonderfully, just waiting for Ex2k to come online.
o POP hangs go back above and repeat to the reboot.
o perfmon open, monitoring pop. nothing that helps.
o install procexp.exe, inetinfo.exe is bogging everything down.
o restart inetinfo.exe (thru procexp.exe, task mgr won't allow)
o all 'protocols' go down... POP is working.. until... SMTP comes online.
o further investigation thru procexp.exe shows that inetinfo.exe controls ALL protocols in the exsysmgr. NNTP, IMAP, POP, SMTP etc.
o SMTP is bogged with spam from the open relay, and grinding inetinfo.exe and the box to a crawl.
o view smtp queue... takes forever to load. Start deleting spam. (this task is a horrid one, no multiple selects.. one at a time.
I could stop smtp and delete the dir storing the info, but that would remove production emails (this was recommened via a Q smc article, however I feel deleting legitimate email traffic is NOT a solution. I could go on and on about some of the other "recommended" "solutions" but that is for another day.
o at 3am I stop deleting messages, restart smtp and will look at it in the morning.
(note: I also limited connect computers to internal block earlier to prevent new connects, this is important if you want to get a handle on it. If you disable relay, but do not limit who can connect, exchange will queue the mail, and then apply the rules..)
o 6am POP3 was responding in a reasonable fashion, still had about 30k spam emails, but it was better. By 11am everything was running smoothly.
Final: The platform is obvisouly Intel (1.4G) with 1Gig Ram. There are about 100+ users on the host. Latest SP's and hotfixes were applied and tested weeks prior.
I haven't had to work NT Admin specific stuff in awhile. Primarly been focusing on Network Security, Unix, NT etc. But I just started working for an Awesome small Colo, so I don't mind getting my hands dirty :-) I was surprised that in two years, the MS support articles seem to have gotten worse. I was surprised as to what MS believed were acceptable solutions to Enterprise level problems. Reboots, downtime, reinstalls. Going thru the MS newsgroups and message forums was worse. The general consensus among admin's seemed that downtime and data loss were acceptable, they just wanted the product to RUN!
oh well, babbled enuff... and if you think I am a Linux Geek or something, just wait till my rant on RH RPM DB corruption :-)