From owner-freebsd-questions Mon Aug 12 07:37:20 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id HAA21962 for questions-outgoing; Mon, 12 Aug 1996 07:37:20 -0700 (PDT) Received: from rk.ios.com (rk.ios.com [198.4.75.55]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id HAA21947 for ; Mon, 12 Aug 1996 07:36:59 -0700 (PDT) Received: (from rashid@localhost) by rk.ios.com (8.7.5/8.7.3) id KAA11353; Mon, 12 Aug 1996 10:35:49 -0400 (EDT) From: Rashid Karimov Message-Id: <199608121435.KAA11353@rk.ios.com> Subject: Re: FreeBSD INN News Server. To: jds@c4systm.com (James D. Stewart) Date: Mon, 12 Aug 1996 10:35:49 -0400 (EDT) Cc: questions@FreeBSD.ORG In-Reply-To: from "James D. Stewart" at Aug 12, 96 03:34:12 am Content-Type: text Sender: owner-questions@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > > The plea... > > We are in process of converting our Usenet News server to FreeBSD. In the > past it has been SCO Unix 3.2.4 (lots of problems), and Redhat Linux > 1.2.13 and 2.0.10 (even more problems). When our satellite feed jumped > from 200+ Mb/day to 800+ Mb/day, we had to do something. We already have > a box with FreeBSD 2.0.5 that has been up constantly for almost a year > with no problems: the only time it has ever had to be rebooted was during > a 5 hour power outage last winter. Other than that, it has been like a > rock! No problems what-so-ever. Is you really wanna do it - go with latest SNAP or the previous one. They've incorporated the latest stability changes in ahc* driver - this is absolutely crucial for server running INND. I'm not sure whether or not the same changes went into 2.1.5. I have Usenet server with 3 2940W SCSIs and 6 HDDs running via CCD and the thing is also very stable - it's up for 2 months serving 200+ news readers in average and doing ~5 incoming/.outgoing feeds at once . It's very fast , even though it'd be better to have separate machinejust to crunch incoming feeds. Rashid > > The News server is running with a AMD586/133 chip (Triton chipset), has > an Adaptec 2940 (their SCSISelect program says aha2940U), and 5 hard > disks with the following layout: > > 1.2 Gb Seagate for /, /var, and /tmp > 1.2 Gb Seagate for /usr, /u, and swap > 2.6 Gb DEC for /var/spool/news (all but alt.*) > 3.9 Gb Fujitsu for /var/spool/news/alt (alt.*) > 540 Mb Connor for /var/spool/news/over.view (overview database) > > When we installed FreeBSD 2.1, we could not get the system to boot - the > book we have said that is probably due to bad disk geometry. All disks > came from existing systems and were operational. We tried using the > manufacturers specs for cyls/hds/sects to no avail. We tried installing > MS-DOS on all drives to no avail. We used the numbers on the existing > FreeBSD system to no avail (it has the same model Seagates). The only way > we got it to boot was to create 1 partition in fdisk and respond No to > the "incompatible with future filesystems" prompt. When we started the > News system, the processing was so slow we could not keep up with the > feed. There are 150 K of articles/day and the best we could do was 1.4 > articles/second. With the Linux setup, we get 3.54 articles/second, and > SCO did 4.22 (different box for SCO). The problems with SCO and Linux are > they need to be rebooted 4 - 6 times a day because of system hang-ups - > mostly general protection faults or bad inodes. With the FreeBSD setup, > the system would just hang - no message and no response to any key combo. > > We know we have this setup wrong somewhere. After all, the other FreeBSD > system here is working wonderfully. We just don't know where. Have you > any hints, ideas, suggestions? Can you point us in the right direction? > > Another problem is that we have 5 disks, and the setup procedure only > installs 4. Is there a way to change that? We could not find a procedure > for adding disk drives after-the-fact in the Install book. Is there a > procedure for that if that is the way to go? > > Any help you can give will be greatly appreciated. The system has been > out of service for 6 days now, and the users are getting really grumpy. > > Thanks a lot! > > Jim Stewart > C4 Systems, Inc. >