[gentoo-user] OT: moving from mod_perl1 to mod_perl2
We've got a pretty complicated middle ware app that we're trying to move from Apache1 w/mod_perl1 to Apache2 w/mod_perl2. The transition is turning into a nice little quagmire. Would anyone like to share some links on getting the most out of compatibility mode, dealing with ModPerl::Registry, or any other nuggets of wisdom? kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] rsync errors
[EMAIL PROTECTED] wrote: Hello, I'm using a server to back up to. It's an rsync server, version 2.6.3. I have approximately 3.9 million files to sync up periodically. The problem is that the client keeps dying on me, and I can't repeat where it happens at, and I don't get any logs server side. All I see on client side is: rsync: connection unexpectedly closed (587704 bytes received so far) [sender] rsync error: error in rsync protocol data stream (code 12) at io.c(359) The machine has a low ammount of ram, however, it has a ton of swap. It goes deep into swap during all of this, but doesn't get anywhere close to using it up. If it was, i'd expect to see kernel messages about such and such being killed off, but I don't. I have other systems doing rsync transfers to other rsync servers, of the same version, with no problems. Granted, none of them are trying to do anywhere close to 3.9 million files. This is the closest explaination I've found. Unfortuately I don't know of a way to make the problem go away. http://www.linuxquestions.org/questions/showthread.php?s=&threadid=265520 kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Host, Schmost!
John Lowell wrote: > I certainly can post ifconfig and will but the netstat -rn is a problem. The command is not available to me with just the basic packages installed. This was a fresh stage1. I'll need to reconfigure /etc/conf.d/net for dhcp service to get the appropriate e-builld. Before I do that, have you any other thoughts or recommendations? Yes I've got a recommendation. Post the ifconfig already. And since you have the ifconfig command I'd suggest posting the netstat -rn immediately after it since it's part of the same package. support root # qpkg -f /sbin/ifconfig sys-apps/net-tools * support root # qpkg -f /bin/netstat sys-apps/net-tools * kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Host, Schmost!
John Lowell wrote: > Thanks for writing. Broadcast with 0 is default in /etc/conf.d/net. I've tried it both ways with no difference being made. Why don't we clear this up and have you post your ifconfig and netstat -rn kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] lost dhcpcd and updatedb/locate
Grant wrote: Since my depcleaning, updatedb and locate are bad commands and dhcpcd is gone. I'm surprised that stuff isn't depended on or part of the system, but minimal is good. dhcpcd is back, but what package is updatedb/locate part of? Should I go for slocate instead? Those are part of the slocate package. -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] removing net-misc/dhcpcd
Antonio Coralles wrote: 'emerge --depclean' suggests removing net-misc/dhcpcd allthough i recieve my ip-address via dhcpd ... So, is it save to remove it ? No. What's happened is that dhcpcd is no longer part of the systems stuff so portage sees it as something that can be removed. I'd emerge dhcpcd directly so that it's known to be part of the system. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] cyrus-sasl-2.1.20 error
death rince wrote: Hi, Well I took the same options ebuild uses when it ran ./configure, i used the same options again. Hence I believe the use flags might be retained. I wouldn't know until it is installed. A wild swing and try. I just want to get over the compile error so that I can compile openldap which has a dependency on this package. Well since it seems to be choking of java, I'd add -java to /etc/portage/package.use for cyrus-sasl assuming you don't need it. I actually have -java in my /etc/make.conf since it tends to have ssues with a number of packages. Ramin -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Rotating cron files
[EMAIL PROTECTED] wrote: I connected to my office LAN a postgresql server with the latest linux gentoo. By means of "crontab" each working day at 3am the following "crono" script - backing up my complete DB and the /etc dir to a network share - is executed: I would like instead to "enumerate" the two files (starting from monday = 1, till friday=5) in order to keep a week backup, such as for instance: curve.1.tar.gz , curve.2.tar.gz, .. curve.5. tar.gz and the same for etc How can I obtain this result with crontab? Here is my db backup script. It's for mysql, but the idea is the same. kashani #!/bin/bash set -x HOST=`hostname --fqdn` LDIR="/var/sqlbackups" MLOGIN="user" MPASS="pass" RHOST="backupserver.domain.com" RDIR="/var/sqlbackups/$HOST" RUSER="root" RKEY="/root/.ssh/dumps" DATE=`date +"%Y%m%d"` OLDDATE=`date --date='6 days ago' +"%Y%m%d"` DUMP="/usr/bin/mysqldump" GZIP="/bin/gzip" SCP="/usr/bin/scp" SSH="/usr/bin/ssh" cd $LDIR $DUMP -u$MLOGIN -p$MPASS --all-databases > $HOST-mysql-fulldump-$DATE $GZIP -9 $HOST-mysql-fulldump-$DATE $SCP -q -i $RKEY $HOST-mysql-fulldump-$DATE.gz [EMAIL PROTECTED]:$RDIR # delete old backups $SSH -i $RKEY -l $RUSER $RHOST rm $RDIR/$HOST-mysql-fulldump-$OLDDATE.gz rm -rf $HOST-mysql-fulldump-$OLDDATE.gz -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Ssh DSA/RSA log in
Digby Tarvin wrote: Seems you are right. I had discovered the need for 'authorized_keys2' some time ago, and been using it ever since. But I just tried moving it to 'authorized_keys' and it appears that is now accepted for protocol 2 also. Yeah authorized_keys2 was a transitional file when protocol 2 first came out IIRC. Mostly as a way to provide backwards compatibility to older clients. It hasn't been needed if you were running current server and client software. In regards to the problem. Check your permissions. .ssh/ and authorized_keys should be owned by the local user and have permissions of 600. If they do not, authentication won't work. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] apache-1.3.33 with mod_perl and mod_php
Felipe de Jesús Molina Bravo wrote: How can install apache-1.3.33 with support for mod_perl and mod_php? My first solution is with variable USE="perl php" is correct? no, apache 1.3 only has two use variables, ssl and pam. emerge apache mod_php mod_perl kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Mbox vs maildir
Nick Rout wrote: > well for a start its way OT. it is not related to gentoo, it is a generic mailserver question, better suited to comp.mail.imap, if you can stomach usenet these days LOL. The past three times mail servers and the like have been mentioned Mr. Rout has remarked how off topic it is. I think that's pretty ridiculous. I'd imagine that 100% of the people on this list interact with email, which is a much better ratio than those that use X, setup a bootsplash, or want sound to work. I quitely delete those threads as being of no interest to me, but interesting and on topic for others on the list. I expect others to give my and a significant portion of the list's interests the same curtesy. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Mbox vs maildir
Neil Bothwick wrote: What is the best way to convert an existing mbox setup to maildir? Reconfiguring the MTA/MDA looks straightforward, but what is the best way to convert existing mailboxes? Most ISPs that I've dealt with either started with .maildir/ or did the switch around '99-00 when the software caught up with the format change. There are a number of scripts that will do the conversion, google/freshmeat. It's fairly straightforward assuming it's a straight spool conversion and not an imap setup where people have lots of separate files. I'd run your conversion script a few times to make sure it's cutting up the spool properly. If you can do it all on a separate server for testing you own account so much the better. Then I'd come up with a plan to change your software to support .maildir/. The MTA side should be just a config change, but it'll messier on the pop/imap side. uw-imap does not support .maildir/ whereas courier-imap only supports it. I think most of the major pop packages can do either these days, but I'd verify that. You'll also need to tweak mutt and pine setting as well if users have access to them. If you have any users that leave their mail on the server, the first day of the switch might be interesting. Chances are their mail client will download all their mail as new. You might be able to tweak this by putting mail into cur/ or renaming files to tell courier that they've been read. Courier adds things to filename to track Sent, Read, etc IIRC. Getting the switch to be completely transparent isn't going to happen, but you can eliminate quite a bit. All in all it's not complicated, but it's going to take some time. If you don't have that many users and haven't had performance issues, I'd probably leave well enough alone. Though doing it now with less users will make your life easier later if you plan to grow the sytem. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Mbox vs maildir
James Colannino wrote: > Also, .maildir has the tendency to fill up your inodes really fast if you have a large scale server, so in some cases where you're dealing with large scale stuff, mbox may be preferable. Actually in large scale stuff you'll need to use shared storage so .maildir/ is still your best choice. You do burn a lot of inodes, but it's a decent trade off against the alternatives. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] linux-2.6.11 is out
Raphael Melo wrote: By the way, I had problems with oracle on kernel 2.6.x. Does that still happens? In my experirence Oracle tends to care more about the version of your C libraries and gcc than the kernel. There are a few posts on the forums about getting the right libs for Oracle under Gentoo. I've current got Oracle 10g running on RHEL v4 which is using kernel 2.6.9. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] OT - dovecot question [WAS OT - uw-imap question]
Michael Sullivan wrote: Will bind-tools give me dig as well? I was afraio I'd have to do an emerge -Ss for it... yes, now install it already!! :) -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Installing Oracle 9iR2
Antoine wrote: There is a guy that works at oracle that uses Gentoo. He has got it (9i, I guess R2) running fine and has instructions somewhere. He may mention something cos I think he reads this list. It certainly doesn't require gcc 2.95. I think I had it completely installed but seeing as I know extremely little about oracle couldn't import the database I wanted to, so couldn't really test it. Yep do a search for oracle and ebrostig on the forums and they should should pop right up. And this fellows site was very helpful for general Oracle knowledge or if you have to go the Redhat route. http://www.puschitz.com/OracleOnLinux.shtml kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Looking up... Connecting... Waiting... Transferring
Grant wrote: I've been optimizing my site's performance by tuning the server-side code and watching how it affects the amount of time Firefox reports as "Waiting". It seems like the "Transferring" time would be optimized by reducing the size of the HTML to download. What about "Looking up" and "Connecting"? I noticed that if I'm doing a lot of clicking around there is pretty much no time spent with either of those, but if I work on something and then come back to the browser after a bit, Firefox can really spend some time there. I'd imagine that is apache's area. What can I do to minimize the time needed to look up and connect? I'm wondering how you're getting any sort of concrete numbers out of anything. If you're eyeballing Firefox and going by feel it's going to be hard for you see where the problem is or if there is one. If it were me, I'd do something like the following: 1. Add a "time to render page" into your test page. I normally see this in php pages, but assume it's not hard to do. This should give you some sort of idea of how complicated the page was to put together o the server side. 2. use curl or other command line tool to pull the page. preferably using the time command. Do this and write times to a file every 30 seconds. 3. Write a script to check server load and record it to a file every 30 secs. 4. send 10 pings or so to the server from the client and record times to a file every 30 secs When you can compare server load to time for the server to render page to time to download the page to ping times you will have interesting data. Otherwise you have no idea if it's the connection, the server or the client. I'm not sure how well most of the what I listed would work in actually implementation, but the idea of getting things put into tools that you can measure in milliseconds is never bad. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] OT - espersunited.com saga; IMAP and POP3 access
Michael Sullivan wrote: Why is it stupid? What is xinetd anyway? What does it do? xinetd is what's often called a super-daemon. It sits there listening for connections on ports you've configured. When connections come in xinetd answers and then calls the correct server to handle it. Once the connection is done, the client exits, xinetd drops the connection, and the server daemon shuts down... not exactly sure on the order for all that, but it doesn't matter much for our purposes. The nice thing about this is that you don't need ten ftp daemons sitting around using RAM while waiting for incoming connections. This helps keep resourvces free on a small system with little RAM that runs many services, but doesn't need to do them all the time. The bad thing is that you have to reread all data and may have some significant start time associated with the daemons. Also you lose any caching of data when the daemon exits. In the case of a busy server that has a dedicated function, xinetd is going to slow things down. Or for real fun try running a database out of xinetd. Back to imap, unless you like the idea of rescanning your entire dir every time you make an imap connection, standalone is better most of the time... for imap/http/mysql. However other daemons with low overhead like pop3/telnet/ftp/tftp would be fine. Here's a bit of link on it, about a third down the page. http://www.squirrelmail.org/wiki/en_US/SquirrelMailPerformance And finally to end with funny inetd story. I once ran ssh 1.2-ish on my Sparc2 out of inetd. It took 30+ seconds to connect each time since ssh had to generate the session keys from scratch, which was no mean feat on a 20mhz chip. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Postfix excessive logging
Maarten wrote: [retrying send since it never made it to the list it seems] Hi List, Today I (more or less) followed some howto's I found in the forum to yield all of postfix (+TLS), amavisd-new, clamav, spamd, courier-imapd-ssl, squirrelmail and apache (+SSL) (but without virtual mail setup). The good news is, all that seems to work fine now. So everything works fine upon first (and second) glance except for the real abundant logging that either postfix or amavis does. I edited all the logging-related options I could find but to no avail. For every single mail I get I see a few hundred(!) lines of logs... :-( I reckon something is running in debug mode, but try as I might I cannot find what is (and where it is configured). A log excerpt is to be found below. There is no postfix config in /etc/conf.d/ and the initscript doesn't set debug mode as far as I could determine. It also seems to me that when I change a loglevel in main.cf that that doesn't change a thing, which leads me to believe that it somehow runs in full debug mode no matter what I do. Look in your /etc/postfix/master.cf for a -v flag in the args field. Remove it and restart. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] OT - espersunited.com saga; IMAP and POP3 access
Michael Sullivan wrote: The problem is that I don't know what's providing IMAP/POP3 on the FC1 box. In redhat-config-services they're just listed as "imap" and "ipop3". I did a full install of FC1 on that box, so all of the packages were already installed; I just had to edit a couple of files in /etc/sysconfig and start the daemons. All I know about the IMAP and POP3 services is that they depend on xinetd. Does that help? What's the output of: rpm -qa | grep -i imap You might also be able to tell what you're running by looking at the xinetd conf files. Though running IMAP out of xinetd has to be one of the stupidest default settings I've ever heard. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] OT - espersunited.com saga; IMAP and POP3 access
Michael Sullivan wrote: Continuing my efforts to recreate the services on my server PC running FC1 on my client PC running Gentoo so that I can install Gentoo on the server PC I am now to the part where I need to enable IMAP and POP3 access on my client PC. I looked on Google for how to do this, and everything I found said that it was for Postfix. I use Sendmail - it's the only MTA I have a book on. Can anyone point me to a free online resource of how to do this with Gentoo and sendmail? I'd install which ever package is providing IMAP/POP3 on your Fedora system. However, If you're using .maildir format emerge courier-imap If you're using mbox format emerge uw-imap /etc/init.d/$service start It really is that easy unless you need to change anything about the default setup, which you shouldn't need to do unless you're building a virtual system. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Filesystem Choice
David Busby wrote: Kashani wrote: ReiserFS does deal with billions of tiny files better than almost all other file systems. You're going to run into inode issues with ext3 as well. However what you're doing sounds like it belongs in a database instead of creating weird data structures within your filesystem. Kashani, Thanks, I searched and found some docs that indicate this too. Your answer was much better than Dave Nebinger's of "search for 'unmaintainable ridiculous software architecture'", that wasn't helpful at all. David, To be honest I had an edit that looked suspiciously like Dave's. I edited because very few people have had a million files/dirs in a file system much less a billion, so I assumed a certain naivety on your part. The idea of using your file system as data structure is tempting because in the beginning it does work and it's simple. After a certain size you start running into really strange problems, backing up becomes next to impossible, you start getting really weird performance, and you use much more space than your 32kb files would lead you to believe. Your design is unworkable. And if you think scanning a db is going to be slow wait till you implement this. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Filesystem Choice
David Busby wrote: I don't want this to turn into one of those ext3 vs xfs conversations. I want to know which filesystem is a better choice for having lots (>4billion) directories. The dirs will be in tree format, so at the root will be 256 dirs, each with 256 sub-dirs, each with 256 sub-dirs. This will go on for 8 to 12 levels deep (I don't know yet) Then each dir will have one small file in it (<32bytes). How would I tune ext3 for this? What about XFS, I read that some of it's features might be useful to me for this project. Any good links? I don't even know what phrases to Google for this one, my results have been unhelpful. ReiserFS does deal with billions of tiny files better than almost all other file systems. You're going to run into inode issues with ext3 as well. However what you're doing sounds like it belongs in a database instead of creating weird data structures within your filesystem. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] samba requires cups?
Covington, Chris wrote: [ebuild N] net-fs/samba-3.0.10 -acl +cups I'd say no, cups isn't required. :) kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] dhcp or dhcpd ?
PK wrote: any idea what I am supposed to do to get it working properly? IIRC there are two main gotchas 1. kernel stuff. Assuming you're using 2.4 you need to have Packet Socket and Socket Filtering compiled in under Networking. If you don't have these I think the dhcp daemon will bitch about it and probably bomb out. 2. missing files for leases. As Nick just posted you'll need to find them and create them and then give them the proper ownership and permissions. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] dhcp or dhcpd ?
Paul Kain wrote: its not in portage ISC DHCP is in portage because that's what actually comprises the dhcp package. So you're already got it installed. * net-misc/dhcp Latest version available: 3.0.1 Latest version installed: [ Not Installed ] Size of downloaded files: 828 kB Homepage:http://www.isc.org/products/DHCP Description: ISC Dynamic Host Configuration Protocol --->->-^^^^ kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] mod_perl 1.27 and apache 1.3
Covington, Chris wrote: Hello, I'm trying to install mod_perl 1.27-r4 on the latest 1.3.33 build of Apache in order to use Request Tracker. While everything emerges properly including rt, I don't have anything on my filesystem named mod_perl.c or mod_perl.so (there's a new libperl.so but that's it) and there's no change to the apache 1.3.33 configuration files to load a mod_perl module. So obviously rt isn't working. I tried adding a LoadModule perl_module /usr/lib/apache-extramodules/libperl.so and LoadModule mod_perl /usr/lib/apache-extramodules/libperl.so to apache.conf but neither works. Apache -v says: Server version: Apache/1.3.33 (Unix) (Gentoo/Linux) Server built: Feb 2 2005 11:43:01 How can I enable mod_perl 1.27-r4? Did you run this command as the mod_perl package instructed? ebuild /var/db/pkg/dev-perl/mod_perl-1.27-r4/mod_perl-1.27-r4.ebuild Then you'll need to add -D PERL to /etc/conf.d/apache and finally stop and start Apache. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] net.eth0
[EMAIL PROTECTED] wrote: hi, in /etc/conf.d/net I have in alias_eth0 : around ~20 aliases alias_eth0="x.x.x.x .. y.y.72.0 y.y.73.0 y.y.74.0 x.x.x.x" it creates these interfaces with netmask 255.255.0.0 and as consequense it inserst a route to y.0.0.0/8 instead of three routes to these 3 networks.. any solution ? You should also have these two lines in your /etc/conf.d/net #broadcast_eth0="192.168.0.255 192.168.0.255" #netmask_eth0="255.255.255.0 255.255.255.0" Uncomment and do them in the same order as your aliases and then your networks will be setup correctly. kashani -- gentoo-user@gentoo.org mailing list
[gentoo-user] OT: console only web browsing
I've always used lynx which seems decent enough, though annoying as I expect any console web browser might be. Are any of them, w3m, links, elinks, amaya, etc better interface and navigation-wise? kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] MySQL related portage recommended *downgrades*?
William Yang wrote: While I understand how to avoid applying these changes, I'm trying to understand why this change is now being recommended. I'm also confused, as http://packages.gentoo.org/search/?sstring=mod_auth_mysql doesn't show mod_auth_mysql as being stable, and the latest stable for mysql is 4.0.22-r2 on x86 architectures. Why these downgrade recommendations? Did a bit more research in the cold and sober light of day. A few things are going on. Looks like in prep for going to 2.8 they changed the versioning somehow. Before 20030510 was the latest. Now it's 1.1 and since that's ancient you've got wacky mysql requirements. I'd mask it, stick with 20030510 or whatever you've installed, and wait for 2.8. 2.8 has a nice little feature at the moment of hijacking all your .htaccess files regardless if they call mysql or not. Instead of having to explicitly call mysql you now have to explicitly deny mysql by adding AuthMySQLEnable Off to .htaccess files that do not use mysql. Hopefully that'll save you from a day or two of insanity. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] mii-tool = ethtool ?
Nick Niemeyer wrote: I am running into problems using mii-tool with the e1000 Intel Cards. Has anyone had any problems using ethtool in place of mii-tool? If so what are things that I should look out for? My understanding is that some cards work for ethtool, some work for mii-diag, some work for both, and some cards don't work at all. I have both installed on my systems, but generally use ethtool as it works fine for the e1000 and the tg3/broadcom stuff which almost all our boxes use. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] MySQL related portage recommended *downgrades*?
William Yang wrote: I've been a Gentoo user for a while now (this is based on a 2004.1 box, but I've been syncing and upgrading every week or two), and I'm a little confused about a recent change. I run mysql 4.0.22-r2 on a production machine on a x86 (i686-pc-linux-gnu); after the last sync of the portage tree, the system now recommends a DOWNGRADE to 3.23.58-r1. # emerge --pretend --update world These are the packages that I would merge, in order: Calculating world dependencies ...done! [ebuild UD] dev-db/mysql-3.23.58-r1 [4.0.22-r2] [ebuild UD] net-www/mod_auth_mysql-1.11 [20030510-r2] While I understand how to avoid applying these changes, I'm trying to understand why this change is now being recommended. I'm also confused, as http://packages.gentoo.org/search/?sstring=mod_auth_mysql doesn't show mod_auth_mysql as being stable, and the latest stable for mysql is 4.0.22-r2 on x86 architectures. Why these downgrade recommendations? Probably a screwup with the Apache herd. They have been reorganizing things, getting rid of mod_contrib, and getting other modules up to date. The actual current version on mod_auth_mysql is 2.8 which I'm using in production. The 1.1 stuff which is really old and the 2003 stuff which is what most people are using. 2.8 should work fine with Mysql 4 or 3 and also with Apache 1 and 2. On my system it wants to upgrade Apache to 2.0 as well. I'd probably keep an eye on any APache stuff for the next couple of weeks as apparently things aren't going to go a smooth as was planned. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] 32BIT/64BIT
William Kenworthy wrote: On a local lug, a statement was made that not a lot of 64 bit software is available for the amd chips yet. With gentoo at least, isnt this largely untrue: if you have the arch set correctly, are you not building 64 bit software, and only 32 bit where 32 bit specific instructions are specified??? An what about the 64 bit binary distros??? Seems to be more a microsoft viewpoint. Hard to say what they were talking about. I'm inclined to think they meant that you can build a 64 bit Linux environment, but most software running would actually use 32 bit. db, compiling, heavy image or music manipulation being the obvious exceptions. Solaris was similar 4-5 years ago. You really had to look for 64 bit software. 64 bit top was essential assuming you were running a 64 bit kernel, but for many other system tools 32 bit was fine as they ran faster without the 64 bit overhead. If I were running a project I'd ask myself if I really needed to be 64 bit. If I'm building a spread sheet program, maybe that answer is yes. If I'm writing a terminal client, maybe that answer is no. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Apache & Mysql Backup System
Seunghyun. Cho wrote: I am running apache/php and Mysql on my gentoo server. I just backup tar copies everyday. When the system is down, it is very difficult to recover it. So I'd like to make automatic backup(or load ballancing) system for my site. How can I implement this on gentoo machine. One is main web&db server. and there is second server sync data in real time. When main server is down, second server should do the job of main server. And data needs to be backuped in realtime(or everyday... real time syncing is much better~) Is anybody implemeted this? Give me some ideas... You've got a few options depending on how much money you have. 1. No shared storage Setup a cron to rsync your Apache dir every X minutes to your second machine. Assuming you're not changing content that often, it would be easier if you had your dev machine, assuming you use one, sync to both servers whenever you push content live. Running frequent rsyncs over large data sets can bog your server down. Mysql replication is incredibly simple these days as long as you're doing it in one direction. Probably take you 30 minute total to set it up. The hard part is switching back to the primary. You don't want that to happen automatically since the primary will not have the latest data. Do some reading on this as you'll really need to know what going on. Run HA scripts or daemons, etc on the two servers. There are a number of packages around that do some simple heartbeat stuff. Again some reading as which one you chooses affects what you can do. 2. Shared storage This requires some significant cash. You hook both servers up to a happy little RAID 5 array. All data is stored on the array. Run your heartbeat scripts and you're pretty much set. You'll need to do a bit more reading about Mysql as you'll NEVER want both Mysql's trying to update the data at the same time. And now the caveats. If you are allowing users to upload pictures and content that doesn't get jammed into the db, doing the first solution is going to suck as you'll need to push the content around pretty quick. Also in the second solution if your storage goes down, you are completely down. HA scripts can get a little funky sometimes. Make sure you're on a switch, with normal ether cards, and have mature drivers. This will go a long way towards making things more stable. Lastly you can always spend a lot more money then this with hardware load balancers, storage with multiple heads, dedicated db servers, etc, but these two options seemed to be closer to what you were looking for. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] RE: little crusade for Gentoo to compile with InnoDB
John Dangler wrote: ABSOLUTELY!!! - If nothing else, add it as a separate entry in Portage (mysql_w_innodb)... I for one, use innodb... Why a separate package when you can add innodb to the USE variables for mysql? emerge -pv mysql dev-db/mysql-4.0.22-r2 [4.0.22] +berkdb -debug +innodb +perl +readline (-selinux) +ssl -static +tcpd 0 kB Yeah it's slightly annoying when you happen to start playing with something that needs innodb, but 9 minutes later you're set. As for innodb being the default storage engine, I'll believe that when the default my.conf starts asking for 5GB for ibdata upfront. :-) kashani -Original Message- From: Arjen Lentz [mailto:[EMAIL PROTECTED] Sent: Friday, January 28, 2005 7:50 AM To: community Subject: little crusade for Gentoo to compile with InnoDB Hi all, Many users run into the problem that on Gentoo, InnoDB is disabled. It's not even compiled in. So, no transactions, no foreign keys. Unless you recompile. Someone reported it as a Gentoo bug: http://bugs.gentoo.org/show_bug.cgi?id=44592 It was closed as "Won't Fix" and the argument was "USE=innodb greatly increases the MySQL compile time. So for those that don't want it, I see no reason to include it." I added a kind note asking for this to be changed/fixed. If you're a Gentoo user, please add your vote to this. And please encourage other Gentoo users to do likewise. Thanks! Regards, Arjen. -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Repeat errors when backing up via rsync to an rsync server
fire-eyes wrote: I'm getting errors when backing up one of my servers to an rsync server using rsync. I am backing up four servers to this one rsync server. Two of them are gentoo. The only one having problems is the one system i'm talking about. On the client side I see this error, after it's gotten well into the backup. It seems to be a random point. rsync: connection unexpectedly closed (5052622 bytes received so far) [sender] rsync error: error in rsync protocol data stream (code 12) at io.c(359) And on the server side: rsync: writefd_unbuffered failed to write 44 bytes: phase "unknown" [generator]: Connection reset by peer (104) The server rsync version is 2.6.3, as is the client. One of the other clients which always succeeds (and has more data to transfer) is also 2.6.3. The other clients are lower versions. Any ideas? I get a similar problem under a different setup. I did find this, which explains what I'm seeing when I rsync from a lightly loaded box to a heavily loaded one. http://www.linuxquestions.org/questions/history/265520 I'm not sure it's applicable in your case, but might be if the problem server has faster disks then the others or is less loaded which might cause it to blow the buffers on the receiving server. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Moving e-mail
Chris Boot wrote: > The actual filename of the emails doesn't have to be the same, as long as it's rougly the same format. My emails look like: 1106132326.25836_0.tarquin.lan:2,RS The 'tarquin.lan' bit varies from host to host, depending on which machine put the mail in the maildir. The first portion is always a different sequence of numbers, if I'm not mistaken it's the received date or something. The last portion is some flags like Seen, Replied, etc... So as long as they are similar courier-imap will understand them. I'll second this. I did a straight-up rsync of my old .maildir/ to the new dir when I switched servers. No problems with imap, webmail, or my mail client. Generally had the same setup on both sides, but there were some minor version and software changes. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] OT: AMD64 vs P4 w/HT
Bob Sanders wrote: There is work going on for the next release of Gentoo on x86_64 in getting the multilib environment working properly. But no, no links on Oracle at the moment. And given everything I've seen on Oracle both when I was actively involved - V3 on a 36-bit Dec KI10, and as a user now, I'd recommend staying away from Oracle if at all possible. Any software that requires a set of specific libraries, even on Red Hat Enterprise, has some serious issues with working at all. Actually general 32bit jails sounded interesting enough on their own since this will likely be a problem for legacy software for the next 1-3 years. Yeah most OSS stuff won't have any issues, but slow moving enterprise "oh you need to pay $5k for the 64 bit version" software will probably be around till the next upgrade cycle. In this case it's one of the 50 largest Oracle installtions in the country... or was when I worked there. I don't think they can avoid Oracle at this point, though moving to amd64/Linux/10g would significantly cut their hardware costs. But if you want to keep track of what's going on with 64-bit and 32-bit multiple library environments, subscribe to - gentoo-amd64. will do. Fiber HBA card? Like Qlogic? It's in the kernel. And any decent FC card has been 64-bit complaint for ages. The only issues I've seen are some Wireless cards and useless winmodems. Real hardware has 64-bit drivers. It's an EMC fiber hba... and of course has vendor supplied drivers that aren't in the kernel at large and aren't 64bit or so they have said. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] OT: AMD64 vs P4 w/HT
Bob Sanders wrote: Yes, it's all 64-bit. It's also possible, with work doing a chroot environment, to run a full 32-bit only user environment. And those things that don't work in 64-bit environment - binary plugins/apps, will eventually get there later this year when Mictosoft gets off its collective butt and ships WinXP64. Put another way - folks like Real have little current incentive when their main customer base remains in 32-bit mode. I know some admins who are looking to try Oracle RAC on amd64, but have some 64bit vs 32bit issues for a number of things. If they could run Oracle in 64bit and some of the helper applications in 32bit that would solve a number of their issues in a production enviroment... testing updates would be much easier. Happen to have any links on this? Also I'd been googling around and hadn't found anything about running 32bit drivers, fiber hba card, in a 64 bit kernel. The answer appears to be no, but no one has said anything definite. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Re: ISP blocking smtp
Gabriel M. Beddingfield wrote: Perhaps you could find/set up an SMTP server that will just relay for you. Set it to listen to port 26 (instead of 25). A friend of mine who runs a small ISP has basically set this up as a permanent solution for his customers who are being blocked by their local ISP's. I'd recommend choosing a port greater than 1024 as some ISP's have specific allow statements for ports less than 1024. It's not a common practice, but it may save you from having to change things later if your ISP decides to go crazy on the security front. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Seeking recommendation for webmail program in portage
[EMAIL PROTECTED] wrote: Can anyone recommend a simple webmail server in portage that uses pop3? Doesn't exist as far I know. Most webmail programs rely on imap to keep things straight on the backend else they'd become pretty complicated. If you want to use mbox, uw-imap is about your only choice. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Virtual mail and mysql
Toby Batch wrote: > I didn't but it doesn't seem to fix it, main.cf: # See also below, section "REJECTING MAIL FOR UNKNOWN LOCAL USERS". # mydestination = $myhostname, localhost.$mydomain, localhost, $transport_maps Gah! My fault on this one. I keep forgetting that I "fixed" a number of things in my installtion. mydestination = $myhostname, localhost.$mydomain, $transport_maps transport_maps = mysql:/etc/postfix/mysql-transport.cf cat /etc/postfix/mysql-transport.cf user= mailsql password= $password dbname = mailsql table = transport select_field= destination where_field = domain hosts = unix:/var/run/mysqld/mysqld.sock This was all marked as optional in the how-to. If this doesn't work, post a copy of your postconf -n. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Virtual mail and mysql
Toby Batch wrote: Jan 13 00:40:19 server12345 postfix/smtpd[8678]: NOQUEUE: reject: RCPT from dyn-81-5-132-21.dsl.eclipse.net.uk[81.5.132.21]: 554 <[EMAIL PROTECTED]>: Relay access denied; from=<[EMAIL PROTECTED]> to=<[EMAIL PROTECTED]> proto=ESMTP helo= Jan 13 00:40:19 server30172 postfix/smtpd[8678]: disconnect from dyn-81-5-132-21.dsl.eclipse.net.uk[81.5.132.21] I think it's likely that Postfix doesn't know that domain123.org.uk is a local destination. Did you add $transport_maps to $mydestination or domain123.uk.org to $mydestination? Either works, but I prefer the former as it'll keep you from having to edit main.cf everytime you add a domain. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Virtual mail and mysql
Correct syntax is the following with puts the y in the right place and adds the / onto the end of .maildir/ INSERT INTO users (id, email, clear, name, uid, gid, homedir, maildir, quota, postfix) values (NULL,'[EMAIL PROTECTED]','changeme','Ringo Starr','1200','1200','/home/vmail','/home/vmail/virt1.com/ringo/.maildir/','','y'); kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] Virtual mail and mysql
Toby Batch wrote: I'm following the virtual mail set guide on the gentoo site and all appears to be going ok. i'm just having a few problem figuring out what should go in each of the mysql tables? I have 5 tables: alias relocated transport users virtual If I have a server called realserver.com with the users john & paul, and two virtual servers, virt1.com with user george and virt2.com with user ringo, then what goes in each table? Assume each user has a unix log on with a home directory in /home and a maildir of $HOME/.maildir I'll assume it's the Postfix how-to. mysql -u root -p# or whatever mysql account has rights use mailsql # or whatever you named it # I'm also going to assume that you used the generic.sql or whatever thing to generate the database. insert into transport (id, domain, destination) values (NULL,'realserver.com','local:'); insert into transport (id, domain, destination) values (NULL,'virt1.com','virtual:'); insert into transport (id, domain, destination) values (NULL,'virt2.com','virtual:'); What we've done so far is said that realserver is local and the other two domains are virtual. Anything designated as local: is going to use system accounts while virtual: needs to do a mysql lookup. BTW the NULL at the beginning is telling mysql to auto-increment the ID field. INSERT INTO users (id, email, clear, name, uid, gid, homedir, maildir, quota, postfix) values (NULL,'[EMAIL PROTECTED]','changeme','George Harrison','1200','1200','/home/vmail','/home/vmail/virt1.com/george/.maildir','y',''); INSERT INTO users (id, email, clear, name, uid, gid, homedir, maildir, quota, postfix) values (NULL,'[EMAIL PROTECTED]','changeme','Ringo Starr','1200','1200','/home/vmail','/home/vmail/virt2.com/ringo/.maildir','y',''); What we have done above is specified our users. Your path, uid, gid, etc might be different. Passwords are going to be clear text unless you've changed the how-to significantly. mkdir -p /home/vmail/virt1.com/george/ mkdir -p /home/vmail/virt2.com/ringo/ chown -R vmail: /home/vmail/ That created dirs and fixed permissions. Postfix will create the .maildir/ when the first mail is received as long as the userdir is there. insert into virtual (id, email, destination) values (NULL,'[EMAIL PROTECTED]','[EMAIL PROTECTED]'); insert into virtual (id, email, destination) values (NULL,'[EMAIL PROTECTED]','[EMAIL PROTECTED],[EMAIL PROTECTED]'); insert into virtual (id, email, destination) values (NULL,'[EMAIL PROTECTED]','[EMAIL PROTECTED]'); The above is how to add some simple aliases. If you decide to add any wild card aliases you must create virtual aliases for real accounts as well. This is because the virtual table gets matched before the users table. And the virtual how-to assumes you'll be using mailman. If you aren't then you can add $transport_maps to $mydestination which will keep you from having to manually add new domains to Postfix One last note, if you want to allow users to manage their own domains once you set them up, check out postfixadmin. It's basically the virtual how-to but with a ton more functionality for managing the system. I'd look into now as the mysql layout is different and trying to cutover later will suck. kashani -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] [OT]: Looking for a CMS..
Daniel G. Siegel wrote: my configuration looks like this: safe_mode=off, GD2 installed php4.3, mysql4.0. I can't use gallery.sf.net, it doesn't work on the server. Thanks for advices. FWIW, I managed to get Gallery working last month under Gentoo, but it was a bit painful. 1. Install Gallery 2. webapp-config Gallery 3. realize that Gallery doesn't like latest imagemagick after an hour of poking it 4. echo ">=media-gfx/imagemagick-6.1" >> /etc/portage/package.mask and emerge again followed by webapp-config 5. spend another hour in a silent fit of rage while it still doesn't work and hoping that the nice coffee girl won't call the cops 6. Realize that it's just a path problem that might have been in the code, caused by trying to fix the imagemagick issue, or screwing around with webapp 7. Move down the street a bit to find something stronger than coffee If you don't need the latest and greatest imagemagick it should work, else you'll need to wait for Gallery 1.5 to be released. You can use the other image kit, NPM or something, but I had problems trying to get it configured as well. kashani -- gentoo-user@gentoo.org mailing list
