Re: MySQL Max connections?
Can anyone tell me if this is the correct way to increase the max number of connections, or how to do it if it's not? I would try to test it myself, but it is a bit hard to do on a busy server and when I would need a huge number of connections. Back in the olden-days (1999) on Solaris boxes, a company I was working at ran into a problem where MySQL hit the maximum number of possible socket descriptors compiled into the kernel, so couldn't get a pipe from the webserver to the DB. Probably not your problem, but keep an eye out for gotchas in the environment too. -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: pop before smtp relay
In the past I have used exact[0] to allow pop-before-relay access, and it has worked well. I'm getting ready to move my mail server and I was looking and it appears that exact is not packaged for Debian I've run exact with Debian/Exim -- and even contributed a little code to support cucipop and some other things. Yes, it is not packaged, but as these things go it is pretty easy to build from source and configure. Yeah, everyone _should_ use AUTH (and heck, SSL), but for a boutique server with few relayers, pop-before-smtp still works very well, thank you, with no existing client configuration changes. -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: pop before smtp relay
In the past I have used exact[0] to allow pop-before-relay access, and it has worked well. I'm getting ready to move my mail server and I was looking and it appears that exact is not packaged for Debian I've run exact with Debian/Exim -- and even contributed a little code to support cucipop and some other things. Yes, it is not packaged, but as these things go it is pretty easy to build from source and configure. Yeah, everyone _should_ use AUTH (and heck, SSL), but for a boutique server with few relayers, pop-before-smtp still works very well, thank you, with no existing client configuration changes. -- Ward
Re: spam from an auto-responder
How can u blame him for some spammer emailing it using ur address as a source? He is the responsible party for mail originated from the pduck.com domain. The minute his auto-responder fired off incorrectly, he became a spammer. When he ignored requests to stop, he became a _willful_ spammer. This is how I can blame him, and why an un-programmable auto-responder is now pretty useless. -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: spam from an auto-responder
How can u blame him for some spammer emailing it using ur address as a source? He is the responsible party for mail originated from the pduck.com domain. The minute his auto-responder fired off incorrectly, he became a spammer. When he ignored requests to stop, he became a _willful_ spammer. This is how I can blame him, and why an un-programmable auto-responder is now pretty useless. -- Ward
Re: spam from an auto-responder
Could someone please help educate this person. You mean the From: header could be forged?! Dear Lord NO! Russell, say it ain't so! I personally like giving forwarding pointers in the 550 text. People can read it, but machines ignore it. (Though I hear Exchange suppresses multi-line 550 text, so be brief!) Am I missing something with this approach? -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: FreeBSD/ Redhat / Debian
At 5:20 PM +1100 1/23/04, Craig Sanders wrote: debian isn't the only linux distribution to have a base system. SLS had one. Slackware had (still has?) one. MCC (if anyone can remember it) had one. these are all dating back to 1993 or 1994, so it's not exactly a new concept in the linux world. Ah Slackware! My first Linux. Such happy memories. (*sniff...*) Maybe I should come home! True, it is not a new concept, but it certainly has been superceded by the whiz-bang gee-whiz throw-everything-in distros here in The New World. i think it's a bogus distinction. The implication he is making is that because there are separate distinct upstream developers for MOST packages(*) in Linux, that means that these packages are not integrated into the system, that the act of packaging is just a quick-and-dirty compile to make a binary. Point taken. The general notion of compile it and throw it over the wall is reductionist and out-of bounds. (In fact, I don't think any act of portage/packaging is ever quick and dirty -- though I wish it were.) this may be true for (some packages in) RH and other distros, but it is certainly not true for Debian. Yes, that's the philosophy we love. BTW, it's not even a true distinction. as you note yourself, the base packages ARE mostly from GNU, and they are as consistent with each other as the equivalents from BSD (but the GNU versions of common tools tend to be vastly superior). Well, I don't know about vastly superior. They're self-consistent, of course, but they are still noisier and have that weird Stallman smell. Though I must say, because they have been grown rather than designed they often meet real-world needs better, no question. Aw, crap, you're right, I like them better too. And I no longer notice the smell. They are superior. Never mind. :-) -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: FreeBSD/ Redhat / Debian
At 5:20 PM +1100 1/23/04, Craig Sanders wrote: debian isn't the only linux distribution to have a base system. SLS had one. Slackware had (still has?) one. MCC (if anyone can remember it) had one. these are all dating back to 1993 or 1994, so it's not exactly a new concept in the linux world. Ah Slackware! My first Linux. Such happy memories. (*sniff...*) Maybe I should come home! True, it is not a new concept, but it certainly has been superceded by the whiz-bang gee-whiz throw-everything-in distros here in The New World. i think it's a bogus distinction. The implication he is making is that because there are separate distinct upstream developers for MOST packages(*) in Linux, that means that these packages are not integrated into the system, that the act of packaging is just a quick-and-dirty compile to make a binary. Point taken. The general notion of compile it and throw it over the wall is reductionist and out-of bounds. (In fact, I don't think any act of portage/packaging is ever quick and dirty -- though I wish it were.) this may be true for (some packages in) RH and other distros, but it is certainly not true for Debian. Yes, that's the philosophy we love. BTW, it's not even a true distinction. as you note yourself, the base packages ARE mostly from GNU, and they are as consistent with each other as the equivalents from BSD (but the GNU versions of common tools tend to be vastly superior). Well, I don't know about vastly superior. They're self-consistent, of course, but they are still noisier and have that weird Stallman smell. Though I must say, because they have been grown rather than designed they often meet real-world needs better, no question. Aw, crap, you're right, I like them better too. And I no longer notice the smell. They are superior. Never mind. :-) -- Ward
Re: FreeBSD/ Redhat / Debian
At 2:14 PM +1100 1/23/04, Craig Sanders wrote: e.g. his long-winded page on the base system, makes it seem as if a base system is something magically distinct that only freebsd has. Linux distributions have had base systems since the early days and, just like *BSD, base system means that it is intended to the base of a system. Yeah? Well, two things: 1) AFAIK, only Debian has a base system that is truly a minimal install. I suppose some other distros do this now too. But a Red Hat install, for instance, is like the circus coming to town -- hardly a base. 2) The larger, more important, point was that the userland components of the FreeBSD base are managed under source code control by the FreeBSD developers and aren't assembled from many places (tho mostly GNU) as Linux distros do. That said, both FreeBSD and Debian share much common philosophy. I guess that's why I like and use both! -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: FreeBSD/ Redhat / Debian
At 2:14 PM +1100 1/23/04, Craig Sanders wrote: e.g. his long-winded page on the base system, makes it seem as if a base system is something magically distinct that only freebsd has. Linux distributions have had base systems since the early days and, just like *BSD, base system means that it is intended to the base of a system. Yeah? Well, two things: 1) AFAIK, only Debian has a base system that is truly a minimal install. I suppose some other distros do this now too. But a Red Hat install, for instance, is like the circus coming to town -- hardly a base. 2) The larger, more important, point was that the userland components of the FreeBSD base are managed under source code control by the FreeBSD developers and aren't assembled from many places (tho mostly GNU) as Linux distros do. That said, both FreeBSD and Debian share much common philosophy. I guess that's why I like and use both! -- Ward
Logrotate weekly prerotate everyday?
Hello Folks:
I call a local script from...
/etc/logrotate.d/apache
...in Debian 3.0 to run Analog reports. It is supposed to run once a
week, but it runs every day:
/var/log/apache/*.log {
weekly
missingok
rotate 52
compress
delaycompress
notifempty
create 640 root adm
sharedscripts
postrotate
/etc/init.d/apache reload /dev/null
endscript
# -- added by ward 28Jul02
prerotate
/etc/run_weekly_analog_reports.sh
endscript
# -- end ward
}
My tiny mind thinks a prerotate block should only be executed
weekly once it has been decided to perform a rotation. Not every
time cron/logrotate peeks into this apache file. What as I missing?
(I have fixed the problem by checking the day of the week in my local
reporting script, but I'd still like to understand my disconnect with
Perfect Understanding of the One True Way(tm).)
Thanks,
-- Ward
Re: exim question
Pete: In your alias file, as your last rule, put *: username Does that really work for you? I had trouble with it because with a line like this, the alias file can never fail. Exim would qualify username and run it through again, it would also run any aliases generated by other rules in the file through a second time, and wind up mapping lots of addresses to username@qualify_domian Maybe I just ran into trouble since I have include_domain on and explicitly handle several virtual domains in my aliases filebut the *: construct was a big enough gun that I sure blew my foot off with it! -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: exim question
Pete: In your alias file, as your last rule, put *: username Does that really work for you? I had trouble with it because with a line like this, the alias file can never fail. Exim would qualify username and run it through again, it would also run any aliases generated by other rules in the file through a second time, and wind up mapping lots of addresses to [EMAIL PROTECTED] Maybe I just ran into trouble since I have include_domain on and explicitly handle several virtual domains in my aliases filebut the *: construct was a big enough gun that I sure blew my foot off with it! -- Ward
Re: exim question
At 6:30 PM -0600 2/20/02, Bernie Berg wrote: im running potato with the unstable packages. How do I get exim to spit all mail that there isn't a user defined for to a certain mail box? so [EMAIL PROTECTED] goes to [EMAIL PROTECTED] 1. An alias file: system_aliases: driver = aliasfile file = /etc/aliases file_transport = address_file search_type= lsearch*@ include_domain = true where /etc/aliases contains: *@domain.com: [EMAIL PROTECTED] 2. A catch all at the end of the director chain (after localuser): catch_all: driver = smartuser domains= domain.com new_address= [EMAIL PROTECTED] -- Ward -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: exim question
At 6:30 PM -0600 2/20/02, Bernie Berg wrote: im running potato with the unstable packages. How do I get exim to spit all mail that there isn't a user defined for to a certain mail box? so [EMAIL PROTECTED] goes to [EMAIL PROTECTED] 1. An alias file: system_aliases: driver = aliasfile file = /etc/aliases file_transport = address_file search_type= lsearch*@ include_domain = true where /etc/aliases contains: [EMAIL PROTECTED]: [EMAIL PROTECTED] 2. A catch all at the end of the director chain (after localuser): catch_all: driver = smartuser domains= domain.com new_address= [EMAIL PROTECTED] -- Ward
