From grahame at angrygoats.net Tue Mar 1 19:58:21 2005 From: grahame at angrygoats.net (Grahame Bowland) Date: Tue Mar 1 19:59:32 2005 Subject: [tech] WAIX Routing ETA In-Reply-To: References: <42121C41.2020505@angrygoats.net> Message-ID: <20050301115821.GA6849@angrygoats.net> On Thu, Feb 17, 2005 at 11:55:34AM +0800, David Adam wrote: > On Tue, 15 Feb 2005, Grahame Bowland wrote: > > > > I don't want to bug them further about this, as they're pretty much > > doing it because they're nice, not because they have to. > > > > Cool. Thanks for the update. Hi guys WAIX routing won't be back for some time. I talked to the guys at AARNET today. They spent a significant amount of time on it yesterday, and have decided that attempting to treat the WAIX only subnet differently is just too painful. They have a bunch of routers in Perth now, not just one.. and several links, and they don't want to put the config everywhere to make sure that the WAIX route doesn't get advertised to the world. What I can do is provide UCC with a list of networks that are WAIX. We can then load that list into the router. The only problem is the potential liability if someone is leaching and WAIX is down... Anyway, committee members: I'm (one) of the the UCC's plants in UCS.. instruct me! - Grahame From davyd at madeley.id.au Tue Mar 1 21:29:52 2005 From: davyd at madeley.id.au (Davyd Madeley) Date: Tue Mar 1 21:31:06 2005 Subject: [tech] WAIX Routing ETA In-Reply-To: <20050301115821.GA6849@angrygoats.net> References: <42121C41.2020505@angrygoats.net> <20050301115821.GA6849@angrygoats.net> Message-ID: <1109683792.9639.6.camel@pingu.madeley.id.au> On Tue, 2005-03-01 at 19:58 +0800, Grahame Bowland wrote: >What I can do is provide UCC with a list of networks that are WAIX. We >can then load that list into the router. The only problem is the >potential liability if someone is leaching and WAIX is down... This would of course give us the advantage of not having to have WAIX ips for machines. Would there be any reliable way to know if WAIX routing is down? The only thing I can think of is checking our next hop past Parnet, and doing it say once every 5 minutes. So the worst that could happen is that we get charged for 5 minutes of extra traffic. If only WAIX appeared at UWA as a separate interface... but alas. Incidently, I was under the impression that we had unfettered access to internet2, but from my quick play the other day, that does not seem to be the case. Can you comment on this? Perhaps we should just wait until the Internet becomes practically free for the entire campus. Is that going to happen at some point this year? -- Davyd Madeley http://www.davyd.id.au/ PGP Fingerprint 08B0 341A 0B9B 08BB 2118 C060 2EDD BB4F 5191 6CDA From davidb-0624 at rcpt.to Tue Mar 1 22:33:30 2005 From: davidb-0624 at rcpt.to (David Basden) Date: Tue Mar 1 22:33:50 2005 Subject: [tech] WAIX Routing ETA In-Reply-To: <1109683792.9639.6.camel@pingu.madeley.id.au> References: <42121C41.2020505@angrygoats.net> <20050301115821.GA6849@angrygoats.net> <1109683792.9639.6.camel@pingu.madeley.id.au> Message-ID: <20050301143330.GA3750@chastity.shikita.com.au> On Tue, Mar 01, 2005 at 09:29:52PM +0800, Davyd Madeley wrote: > On Tue, 2005-03-01 at 19:58 +0800, Grahame Bowland wrote: > > >What I can do is provide UCC with a list of networks that are WAIX. We > >can then load that list into the router. The only problem is the > >potential liability if someone is leaching and WAIX is down... > > This would of course give us the advantage of not having to have WAIX > ips for machines. Would there be any reliable way to know if WAIX > routing is down? It's not too hard to see if the routes are being advertised to WAIX at any point in time. The thing is that you want to make sure that other WAIX peers are going to be honouring that path to get to WAIX. If a theoretical WAIX peer 'DodgyNet' messed up it's router config, and ignored the advertisements from WAIX, everything could seem fine (routes being advertised, outbound traffic from UCC goes via WAIX), but the data going back to UCC could be via a charged link. It's not a huge problem, unless of course someone is mirroring DeadRat Linux from DodgyNet, and UCC gets charged by the meg. Low chance, but high cost. David From zanchey at ucc.gu.uwa.edu.au Tue Mar 1 22:34:52 2005 From: zanchey at ucc.gu.uwa.edu.au (David Adam) Date: Tue Mar 1 22:35:33 2005 Subject: [tech] Plone dead? Message-ID: Is anyone doing the move from beta. to www. ? The Plone site is returning a 500. I'm trying to sign up a new member, so can someone poke it please? Cheers, David Adam --- zanchey@ucc.gu.uwa.edu.au Medicine: And you thought hacking computers was complex. From paul at bur.st Tue Mar 1 22:41:11 2005 From: paul at bur.st (Paul Day) Date: Tue Mar 1 22:41:28 2005 Subject: [tech] WAIX Routing ETA In-Reply-To: <20050301143330.GA3750@chastity.shikita.com.au> References: <42121C41.2020505@angrygoats.net> <20050301115821.GA6849@angrygoats.net> <1109683792.9639.6.camel@pingu.madeley.id.au> <20050301143330.GA3750@chastity.shikita.com.au> Message-ID: On Tue, 1 Mar 2005, David Basden wrote: > It's not too hard to see if the routes are being advertised to WAIX at > any point in time. The thing is that you want to make sure that other > WAIX peers are going to be honouring that path to get to WAIX. If a > theoretical WAIX peer 'DodgyNet' messed up it's router config, and > ignored the advertisements from WAIX, everything could seem fine (routes > being advertised, outbound traffic from UCC goes via WAIX), but the data > going back to UCC could be via a charged link. Unlike most, I thought parnet classed incoming traffic as free or charged based on its outbound route, rather than its inbound route? PD -- Paul Day Web: www.bur.st/~paul GPG Key ID: 7FF655A8 From davyd at madeley.id.au Tue Mar 1 22:46:42 2005 From: davyd at madeley.id.au (Davyd Madeley) Date: Tue Mar 1 22:47:50 2005 Subject: [tech] Plone dead? In-Reply-To: References: Message-ID: <1109688402.9639.9.camel@pingu.madeley.id.au> On Tue, 2005-03-01 at 22:34 +0800, David Adam wrote: >Is anyone doing the move from beta. to www. ? The Plone site is returning >a 500. Something has gone critically wrong! Bernard is looking into it, but hopefully we didn't loose the unbacked-up-binary-database due to disk corruption. In case there is something seriously wrong, Bernard has made a copy of the disk image. If we lose the new website, on the day I was hoping to cut it over, I will actually, honestly cry. Really, I mean it. Cry like a little girl! --d (nail biting) -- Davyd Madeley http://www.davyd.id.au/ PGP Fingerprint 08B0 341A 0B9B 08BB 2118 C060 2EDD BB4F 5191 6CDA From zanchey at ucc.gu.uwa.edu.au Tue Mar 1 22:52:41 2005 From: zanchey at ucc.gu.uwa.edu.au (David Adam) Date: Tue Mar 1 22:52:48 2005 Subject: [tech] Plone dead? In-Reply-To: <1109688402.9639.9.camel@pingu.madeley.id.au> References: <1109688402.9639.9.camel@pingu.madeley.id.au> Message-ID: Uncool. David. From bernard at blackham.com.au Wed Mar 2 00:22:54 2005 From: bernard at blackham.com.au (Bernard Blackham) Date: Wed Mar 2 00:23:12 2005 Subject: [tech] Plone dead? In-Reply-To: References: <1109688402.9639.9.camel@pingu.madeley.id.au> Message-ID: <20050301162254.GA5114@blackham.com.au> On Tue, Mar 01, 2005 at 10:52:41PM +0800, David Adam wrote: > Uncool. Indeed. Well the current status is that we can't recover the database in one piece. We're currently in the slow process of extracting the useful bits (namely the HTML and CSS content). The current suspects for this are either (a) some unclean shutdowns earlier today, or (b) Xen which is now running on mussel (neither of which should've caused something this catastrophic though). I'm somewhat baffled and still a little dubious as to the consistency of mussel's filesystems - I'll take it down and do a full fsck tomorrow arvo. Bernard. -- Bernard Blackham From bernard at blackham.com.au Thu Mar 3 01:19:44 2005 From: bernard at blackham.com.au (Bernard Blackham) Date: Thu Mar 3 01:20:02 2005 Subject: [tech] Downtime: Decomissioning meito Thursday arvo Message-ID: <20050302171944.GA26605@blackham.com.au> meito will be decomissioned today and its services replaced by manbo. /away is currently unmounted most places whilst the copying runs overnight, and won't be accessible until the move is complete. Things may start breaking sometime in the afternoon; namely, the website and user pages, thinterms, Windows logons. Service should be restored shortly thereafter. Bernard. -- Bernard Blackham From adrian at ucc.gu.uwa.edu.au Thu Mar 3 09:51:09 2005 From: adrian at ucc.gu.uwa.edu.au (Adrian Chadd) Date: Thu Mar 3 09:51:17 2005 Subject: [tech] WAIX Routing ETA In-Reply-To: References: <42121C41.2020505@angrygoats.net> <20050301115821.GA6849@angrygoats.net> <1109683792.9639.6.camel@pingu.madeley.id.au> <20050301143330.GA3750@chastity.shikita.com.au> Message-ID: <20050303015108.GB10912@ucc.gu.uwa.edu.au> On Tue, Mar 01, 2005, Paul Day wrote: > >theoretical WAIX peer 'DodgyNet' messed up it's router config, and > >ignored the advertisements from WAIX, everything could seem fine (routes > >being advertised, outbound traffic from UCC goes via WAIX), but the data > >going back to UCC could be via a charged link. > > Unlike most, I thought parnet classed incoming traffic as free or charged > based on its outbound route, rather than its inbound route? Its all a bit strange. Specifically, yes, Parnet bill us on traffic based on the destination route they have for the given subnet. This makes the rather evil assumption that traffic is symmetric. It isn't /always/ symmetric, as Paul has found out from time to time :), but the amount of traffic under normal circumstances that is mis billed is almost always under the noise floor[1]. Answering Davyd: yes, the traffic from internet-2 and whatnot is 'free', but here's the catch - most of it comes in the same interface (Parnet) as the bulk of our traffic. The only time this isn't the case is if it comes in via our link to GrangeNet. The reason that traffic over GrangeNet isn't free is because the internet isn't a stable place and we don't want/need people whinging at us because traffic is free one day and isn't free the next. People /will/ want to start transferring huge, huge bulk amounts of traffic to and from free sites. Yes, legitimate stuff. There's solutions to this but it would require re-engineering of the core. I won't go into examples, I'm sure the Clueful out there know what could be done with a network full of Cisco switch/routers. And that won't happen just to make this 'free' network stuff work. We want it, but it'll be part of a general re-engineering when we start shoe-horning in Quality of Service(tm). In short: I'm sorry guys, but as much as I'd love the university to be able to tell that traffic is indeed free when we bloody get it and be able to communicate this effectively out to our edge devices which also do filtering, its Just Not Easily Possible Right Now Without Exposing Us To Lots Of Shit, and therefore won't happen unless Grahame, Toivo or I invent some 5 minute hack that'll last 3 months before breaking. See if you can simply come up with an alternative. :) Adrian [1] Noise floor: defined (by me) as the constant chatter you hear on todays internet by viruses, blackhat scans, stupidly configured DNS and other such tripe. From frenchie at frenchie.id.au Fri Mar 4 10:34:41 2005 From: frenchie at frenchie.id.au (James French) Date: Fri Mar 4 10:34:52 2005 Subject: [tech] Azure rejecting logins Message-ID: <4227C941.5070108@frenchie.id.au> Azure is rejecting ssh logins at the moment. It would seem that it doesn't have the home dirs mounted as its asking me for a password rather than doing public-key authentication. I haven't checked a login at the machine itself. -- James French: frenchie@frenchie.id.au http://www.frenchie.id.au From trs80 at ucc.gu.uwa.edu.au Fri Mar 4 13:59:46 2005 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Fri Mar 4 13:59:53 2005 Subject: [tech] Azure rejecting logins In-Reply-To: <4227C941.5070108@frenchie.id.au> Message-ID: On Fri, 4 Mar 2005, James French wrote: > Azure is rejecting ssh logins at the moment. It would seem that it > doesn't have the home dirs mounted as its asking me for a password > rather than doing public-key authentication. piggery was off the network for some reason, which provides nis for azure. Bernard kindly rebooted it, and now it's working ok. /home wasn't mounted either, but that was a side issue. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From jahanshi at absolutemotion.com Fri Mar 4 13:48:07 2005 From: jahanshi at absolutemotion.com (Paul A.Davis) Date: Fri Mar 4 14:13:33 2005 Subject: [tech] [SPAM] Cialis - wholesale price Message-ID: <5ad801c5207d$2a423608$bb6b7cfc@absolutemotion.com> Skipped content of type multipart/alternative From trs80 at ucc.gu.uwa.edu.au Tue Mar 8 19:14:52 2005 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Tue Mar 8 19:15:00 2005 Subject: [tech] morwong's back Message-ID: As you've probably noticed, morwong is back. It stopped responding about midday, then failed to POST upon rebooted. I pulled it out and reseated all the ram and the scsi card, and it seems to be ok now. In other words, it just wanted a cup of tea, a Bex and a good lie down. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From zanchey at ucc.gu.uwa.edu.au Thu Mar 10 11:06:32 2005 From: zanchey at ucc.gu.uwa.edu.au (David Adam) Date: Thu Mar 10 11:06:45 2005 Subject: [tech] Nautilus config Message-ID: Can somene from wheel or someone else sufficiently enlightened, empowered and enthusiastic check out Nautilus, please? /bin/bash doesn't exist, making it a bit difficult to get a command shell (Terminal won't let you start commands without a valid shell, or something) Also, when I login, my home directory won't show up in the Finder ("Access denied"-style messages). I know almost no-one uses the old Mac now, but I have recently re-introduced the joys of Bolo to some of our younger members, and of cousre XBolo is the flavour of the month. Cheers, David Adam zanchey@ucc.gu.uwa.edu.au From alastair at ucc.gu.uwa.edu.au Fri Mar 11 19:16:08 2005 From: alastair at ucc.gu.uwa.edu.au (Alastair Irvine) Date: Fri Mar 11 19:16:57 2005 Subject: [tech] cybium linux problem Message-ID: <20050311111608.GA121988@morwong.ucc.gu.uwa.edu.au> When attempting to boot to Linux* on cybium, it gives a kernel panic after mounting the filesystems; apparently something "tried to kill init". * 2.4 kernel; distro uknown -- ... The only certainty is that nothing is certain. _____________________________________________________________________ | | | -=*Alastair Irvine*=- | | C-monkey/wanderer/board&RPGer/net-nut alastair@ucc.gu.uwa.edu.au | |_____________________________________________________________________| From davidb-0624 at rcpt.to Fri Mar 11 19:24:23 2005 From: davidb-0624 at rcpt.to (David Basden) Date: Fri Mar 11 19:49:32 2005 Subject: [tech] cybium linux problem In-Reply-To: <20050311111608.GA121988@morwong.ucc.gu.uwa.edu.au> References: <20050311111608.GA121988@morwong.ucc.gu.uwa.edu.au> Message-ID: <20050311112423.GH8659@chastity.shikita.com.au> On Fri, Mar 11, 2005 at 07:16:08PM +0800, Alastair Irvine wrote: > When attempting to boot to Linux* on cybium, it gives a kernel panic after > mounting the filesystems; apparently something "tried to kill init". > > * 2.4 kernel; distro uknown Is this the default install? If so, it's Debian... I'll take a look at it next time i'm in (assuming someone hasn't fixed it already), but "tried to kill init" normally means that the first process started by the kernel (almost always init, but can also be something in the initrd) exited, which doesn't normally happening[0]. Normally the stuff above might show what happened before that, to cause init (or more likely, initrd setting up the system. man initrd on a linux box for more info) to die. David [0] From typical UNIX IPC stuff, if you kill off init (pid 1), all children of init are killed off, which is pretty much everything. The Linux kernel realises that there (probably) isn't anything running in userspace anymore, and points out that the computer probably isn't going to be doing anything very interesting from that point onward. From trs80 at ucc.gu.uwa.edu.au Fri Mar 11 21:36:17 2005 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Fri Mar 11 21:36:35 2005 Subject: [tech] cybium linux problem In-Reply-To: <20050311112423.GH8659@chastity.shikita.com.au> Message-ID: On Fri, 11 Mar 2005, David Basden wrote: > On Fri, Mar 11, 2005 at 07:16:08PM +0800, Alastair Irvine wrote: > > When attempting to boot to Linux* on cybium, it gives a kernel panic after > > mounting the filesystems; apparently something "tried to kill init". > > > > * 2.4 kernel; distro uknown > > Is this the default install? If so, it's Debian... > > I'll take a look at it next time i'm in (assuming someone hasn't fixed > it already), but "tried to kill init" normally means that the first > process started by the kernel (almost always init, but can also be > something in the initrd) exited, which doesn't normally happening[0]. > > Normally the stuff above might show what happened before that, to > cause init (or more likely, initrd setting up the system. man initrd on > a linux box for more info) to die. /sbin/init is AWOL, as is /bin/sh. Nothing to do with an initrd. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From matt at ucc.asn.au Sat Mar 12 16:10:06 2005 From: matt at ucc.asn.au (Matt Johnston) Date: Sat Mar 12 16:10:11 2005 Subject: [tech] morwong's back In-Reply-To: References: Message-ID: <20050312081006.GA2772@morwong.ucc.gu.uwa.edu.au> On Tue, Mar 08, 2005 at 07:14:52PM +0800, James Andrewartha wrote: > As you've probably noticed, morwong is back. It stopped responding about > midday, then failed to POST upon rebooted. I pulled it out and reseated > all the ram and the scsi card, and it seems to be ok now. In other words, > it just wanted a cup of tea, a Bex and a good lie down. The same thing happened today. I didn't bother dismantling things, just turned the power off for 1/2 an hour and repowered it, things came back. Mysterious. Matt From elixxir at ucc.asn.au Wed Mar 16 15:44:33 2005 From: elixxir at ucc.asn.au (Paul Marinceu) Date: Wed Mar 16 15:44:59 2005 Subject: [tech] KVM Gear Message-ID: <4237E3E1.6010808@ucc.asn.au> Hey, I've been looking at various low to mid-end KVM switches and just wanted to see if people have had some experience with certain brands and whether you want to recommend some more than others. This is to manage 4 rack servers using PS/2 and VGA. I've had a look at the following: D-Link DKVM-4K - 4 Port Mini KVM Switch with build-in cables Belkin OmniView? SOHO 4-Port Switch - PS/2 only These both retail under $200 and seem reasonable. What do people think? --p From bernard at blackham.com.au Wed Mar 23 00:19:44 2005 From: bernard at blackham.com.au (Bernard Blackham) Date: Wed Mar 23 00:20:53 2005 Subject: [tech] IPv6 happier Message-ID: <20050322161944.GF4219@blackham.com.au> We discovered an extra spurious IPv6 route for UCC that was making all sorts of things unhappy. This has been removed, and IPv6 at UCC has since picked up. I've created the zones ipv6.ucc.{asn,gu{ild,}.uwa.edu}.au to contain only IPv6 addresses, to avoid issues with services that don't bind on v6. The zone file is currently being updated by hand, as I couldn't find the urge to hack up zonemake.py. The machines that I've confirmed have working[1] and firewalled v6 access are: * mussel * martello * manbo * madako (not for general members) * mooneye (not for general members) * flying (not for general members, but should forward 11b wireless clients happily - haven't confirmed that yet). This does mean that for people that have no free connectivity to UCC because UCC WAIX[2] is down, you can now connect over IPv6 from anywhere in the world for free and without tunneling![3] Bernard. [1] They work at the moment. They should work after a reboot. I haven't tested this. [2] You know what I mean. [3] Well, UCS does that for us. Or you might decide to do your own tunnel to get at the IPv4 internet over IPv6... -- Bernard Blackham From grahame at angrygoats.net Wed Mar 23 01:37:16 2005 From: grahame at angrygoats.net (Grahame Bowland) Date: Wed Mar 23 01:40:11 2005 Subject: [tech] IPv6 happier In-Reply-To: <20050322161944.GF4219@blackham.com.au> References: <20050322161944.GF4219@blackham.com.au> Message-ID: <20050322173716.GA26759@angrygoats.net> On Wed, Mar 23, 2005 at 12:19:44AM +0800, Bernard Blackham wrote: > We discovered an extra spurious IPv6 route for UCC that was making > all sorts of things unhappy. This has been removed, and IPv6 at UCC > has since picked up. I've created the zones > ipv6.ucc.{asn,gu{ild,}.uwa.edu}.au to contain only IPv6 addresses, > to avoid issues with services that don't bind on v6. > > The zone file is currently being updated by hand, as I couldn't find > the urge to hack up zonemake.py. The machines that I've confirmed > have working[1] and firewalled v6 access are: > > * mussel > * martello > * manbo > * madako (not for general members) > * mooneye (not for general members) > * flying (not for general members, but should forward 11b wireless > clients happily - haven't confirmed that yet). > > This does mean that for people that have no free connectivity to UCC > because UCC WAIX[2] is down, you can now connect over IPv6 from > anywhere in the world for free and without tunneling![3] > > Bernard. > > [1] They work at the moment. They should work after a reboot. I > haven't tested this. > [2] You know what I mean. > [3] Well, UCS does that for us. Or you might decide to do your own > tunnel to get at the IPv4 internet over IPv6... All credit to Bernard for spotting the dumb routing error causing the problem :) I'm going to be getting UWA onto a native IPv6 link that comes over the same connection as our IPv4, rather than the slow and kind of fragile tunnel over IPv4 to Sydney. I'll chase this up soon :-) From trs80 at ucc.gu.uwa.edu.au Wed Mar 23 18:41:43 2005 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Wed Mar 23 18:41:57 2005 Subject: [tech] martello speed Message-ID: The new SATA controller has been installed in martello, and I've been benchmarking various RAID configurations with bonnie++. The results are somewhat interesting (commands used to set up each configuration included in case I did something stupid): RAID 5 ------ mdadm -Cv -l5 -n4 -c128 /dev/md0 /dev/sx8/[0123]p1 (this only resynced at 22MB/s) pvcreate /dev/md0 vgcreate reliable /dev/md0 lvcreate -L 40G -n home reliable mke2fs -b 4096 -E stride=32 -j -m 1 -O dir_index -v /dev/mapper/reliable-home bonnie++ -d bonnie++ -u trs80 -m martello Version 1.03 ------Sequential Output------ --Sequential Input- --Random- -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP martello 4G 38175 92 65903 27 20999 8 26989 60 58108 12 170.9 0 (fiddling with blockdev --setra didn't do much) RAID 1+0 -------- mdadm -Cv -l1 -n2 /dev/md0 /dev/sx8/[01]p1 mdadm -Cv -l1 -n2 /dev/md1 /dev/sx8/[23]p1 (these would only resync at a total rate of 40MB/s, regardless of how whether I stopped one array or not. Further, when only md0 was resyncing, an hdparm of /dev/sx8/3p1 returned 30MB/s, compared to the normal 55MB/s) pvcreate /dev/md0 /dev/md1 vgcreate reliable /dev/md0 /dev/md1 lvcreate -L 40G -i2 -I32 -n home reliable mke2fs -b 4096 -E stride=8 -j -m 1 -O dir_index -v /dev/mapper/reliable-home bonnie++ -d bonnie++ -u trs80 -m martello Version 1.03 ------Sequential Output------ --Sequential Input- --Random- -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP martello 4G 39478 97 66125 29 19881 5 29657 65 58542 8 199.8 0 Bare disk --------- mke2fs -j -m 1 -O dir_index -v /dev/sx8/0p1 bonnie++ -d bonnie++ -u trs80 -m martello Version 1.03 ------Sequential Output------ --Sequential Input- --Random- -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP martello 4G 38044 93 57773 25 17714 6 27826 60 45065 5 215.6 0 The bare disk result bears out the hdparm value of 55MB/s. The question is, where's the performance improvement that should happend for RAID 1+0 and RAID 5? The slow resync and hdparm result during array construction says to me that the ports on the controller are not independent. A little searching finds http://forums.storagereview.net/index.php?showtopic=17794&hl=promise which claims "The linux driver sends only one command to the board at a time." -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From grahame at angrygoats.net Thu Mar 24 00:39:55 2005 From: grahame at angrygoats.net (Grahame Bowland) Date: Thu Mar 24 00:44:13 2005 Subject: [tech] WAIX and Internet2 access working Message-ID: <20050323163955.GA2083@angrygoats.net> Hi guys I've set up code to generate a comprehensive list of all networks that should be free for a given 24 hour period. Before I go any further, I'd just like to explicitly state that this could break, and it could cost UCC money. So we need to make sure we don't mess up and keep an eye on things for a few days. If something goes wrong, blame me with my UCC hat on, not me with my UCS hat on :) There is a large iptables chain FREENETS on madako. This allows networks to be matched based on whether they are reached via Grangenet or the various non-byte charged bits of AARNET. I've set things up so that at boot time, hosts that are limited to FREENETS have: access to all of FREENETS (-A FORWARD -d 130.95.13.18 -j FREENETS) no access to anything else (-A FORWARD -d 130.95.13.18 -j DROP) FREENETS is initialised to just 130.95/16 at boot time. FREENETS is updated overnight via script which can be found in root's crontab on madako. AARNET update their information at 8:30pm, I update UCS's information at 9pm to be safe. UCC retrieves the processed data at 9:05pm. I've modified the AARNET-IP-IN access list on villa so that mussel is unfiltered. All filtering has to be done on madako, and if madako decides to let mussel get to charged hosts that's just too bad.. but it seems to be fine now. At the moment everything seems fine. I'd appreciate other people checking over what I've done. There are approximately 20,000 prefixes reachable from mussel now, with no theoretical charge to UCC :-) mussel::~ $ ping www.gnome.org PING www.gnome.org (12.107.209.247) 56(84) bytes of data. 64 bytes from window.gnome.org (12.107.209.247): icmp_seq=1 64 bytes from window.gnome.org (12.107.209.247): icmp_seq=2 Anyway, test it out and let me know how you go. It would be cool if someone could write some stuff to check if charged hosts are reachable, and bleep loudly if they are. This might save our bacon sometime in the future :) Granting further hosts access to the FREENETS prefixes requires a change on the UCS router, so give me a poke if that needs doing :) We know things are probably free for 24 hours because AARNET base their billing for a 24 hour period off the same data we're generating our iptables chain from. It's not likely things changing during the day within AARNET will cause a problem. From elixxir at ucc.asn.au Thu Mar 24 10:40:46 2005 From: elixxir at ucc.asn.au (Paul Marinceu) Date: Thu Mar 24 10:41:01 2005 Subject: [tech] WAIX and Internet2 access working In-Reply-To: <20050323163955.GA2083@angrygoats.net> References: <20050323163955.GA2083@angrygoats.net> Message-ID: <424228AE.5040701@ucc.asn.au> Grahame Bowland wrote: >Hi guys > >I've set up code to generate a comprehensive list of all networks that >should be free for a given 24 hour period. > > Kudos Grahame! That's awesome :) But what about inbound? Pinging 203.24.97.228 with 32 bytes of data: Request timed out. Ping statistics for 203.24.97.228: Packets: Sent = 1, Received = 0, Lost = 1 (100% loss), Control-C From matt at ucc.asn.au Thu Mar 24 10:49:56 2005 From: matt at ucc.asn.au (Matt Johnston) Date: Thu Mar 24 10:50:06 2005 Subject: [tech] WAIX and Internet2 access working In-Reply-To: <424228AE.5040701@ucc.asn.au> References: <20050323163955.GA2083@angrygoats.net> <424228AE.5040701@ucc.asn.au> Message-ID: <20050324024955.GH268881@morwong.ucc.gu.uwa.edu.au> On Thu, Mar 24, 2005 at 10:40:46AM +0800, Paul Marinceu wrote: > Grahame Bowland wrote: > > >Hi guys > > > >I've set up code to generate a comprehensive list of all networks that > >should be free for a given 24 hour period. > > > > > Kudos Grahame! That's awesome :) > But what about inbound? > > Pinging 203.24.97.228 with 32 bytes of data: The 203 IPs don't exist any more. Just use mussel's normal IP and it should work... Matt From davyd at madeley.id.au Thu Mar 24 11:32:30 2005 From: davyd at madeley.id.au (Davyd Madeley) Date: Thu Mar 24 11:32:56 2005 Subject: [tech] WAIX and Internet2 access working In-Reply-To: <20050324024955.GH268881@morwong.ucc.gu.uwa.edu.au> References: <20050323163955.GA2083@angrygoats.net> <424228AE.5040701@ucc.asn.au> <20050324024955.GH268881@morwong.ucc.gu.uwa.edu.au> Message-ID: <1111635150.13174.58.camel@pingu.madeley.id.au> On Thu, 2005-03-24 at 10:49 +0800, Matt Johnston wrote: > > > > Pinging 203.24.97.228 with 32 bytes of data: > > The 203 IPs don't exist any more. It is the sanity we dreamed of, plus 20% more of the internet! Everyone should rub Grahame's head, or something. --d -- Davyd Madeley http://www.davyd.id.au/ PGP Fingerprint 08B0 341A 0B9B 08BB 2118 C060 2EDD BB4F 5191 6CDA From grahame at angrygoats.net Thu Mar 24 12:20:00 2005 From: grahame at angrygoats.net (Grahame Bowland) Date: Thu Mar 24 12:23:01 2005 Subject: [tech] WAIX and Internet2 access working In-Reply-To: <20050324024955.GH268881@morwong.ucc.gu.uwa.edu.au> References: <20050323163955.GA2083@angrygoats.net> <424228AE.5040701@ucc.asn.au> <20050324024955.GH268881@morwong.ucc.gu.uwa.edu.au> Message-ID: <20050324042000.GA5809@angrygoats.net> On Thu, Mar 24, 2005 at 10:49:56AM +0800, Matt Johnston wrote: > On Thu, Mar 24, 2005 at 10:40:46AM +0800, Paul Marinceu wrote: > > Grahame Bowland wrote: > > > > >Hi guys > > > > > >I've set up code to generate a comprehensive list of all networks that > > >should be free for a given 24 hour period. > > > > > > > > Kudos Grahame! That's awesome :) > > But what about inbound? > > > > Pinging 203.24.97.228 with 32 bytes of data: > > The 203 IPs don't exist any more. > > Just use mussel's normal IP and it should work... Can I stop routing the 203 IPs about campus? We should delete them off the hosts before I do this, I'm guessing.. I can be about tonight and lend a hand doing that. From elixxir at ucc.asn.au Thu Mar 24 13:44:02 2005 From: elixxir at ucc.asn.au (Paul Marinceu) Date: Thu Mar 24 13:44:17 2005 Subject: [tech] WAIX and Internet2 access working In-Reply-To: <1111635150.13174.58.camel@pingu.madeley.id.au> References: <20050323163955.GA2083@angrygoats.net> <424228AE.5040701@ucc.asn.au> <20050324024955.GH268881@morwong.ucc.gu.uwa.edu.au> <1111635150.13174.58.camel@pingu.madeley.id.au> Message-ID: <424253A2.7040807@ucc.asn.au> Davyd Madeley wrote: > On Thu, 2005-03-24 at 10:49 +0800, Matt Johnston wrote: > >>>Pinging 203.24.97.228 with 32 bytes of data: >> >>The 203 IPs don't exist any more. > > > It is the sanity we dreamed of, plus 20% more of the internet! > > Everyone should rub Grahame's head, or something. Ahh. Nice, it feels just like if UCC's been disconnected from the rest of the net all this time and is now finally back (and then some) Woo! ;) If this lasts, I'll personally buy Grahame a pint of beer[*]. --p [*] Subject to the fact that he reminds me of this some time in the future when I'll have forgotten. From castiglione at mac.com Thu Mar 24 14:43:03 2005 From: castiglione at mac.com (Thomas Castiglione) Date: Thu Mar 24 14:43:28 2005 Subject: [tech] WAIX and Internet2 access working In-Reply-To: <20050323163955.GA2083@angrygoats.net> References: <20050323163955.GA2083@angrygoats.net> Message-ID: <78e90921d02f3501d08a704c42d9132b@mac.com> This is awesome news. Thank you Grahame! From matt at ucc.asn.au Sun Mar 27 17:34:04 2005 From: matt at ucc.asn.au (Matt Johnston) Date: Sun Mar 27 18:00:08 2005 Subject: [tech] mail on mussel In-Reply-To: <20050210145124.GA15757@ucc.gu.uwa.edu.au> References: <20050210145124.GA15757@ucc.gu.uwa.edu.au> Message-ID: <20050327093404.GA520064@morwong.ucc.gu.uwa.edu.au> On Thu, Feb 10, 2005 at 10:51:24PM +0800, Alastair Irvine wrote: > Hi, all. Can somebody please change mussel:/etc/mailname so that it > contains: > > ucc.gu.uwa.edu.au > > (this filename exists as a string in the mutt binary). The problem was that mussel's exim config was wrong - I just fixed it now (qualify_domain was set wrong). Hopefully now mail sent from mussel won't say "user@mussel.ucc...". Previous attempts to fix the problem by editing /etc/postfix/main.cf directory failed.... > A while ago someone mentioned that e-mailing from mussel was not a good > idea. Any further info? atimes don't get updated on mussel (this is a problem with morwong's NFS), so it will report new mail in dirs which don't have any mail. Or perhaps it's mtimes - I forget. If you don't notice the problem then it should be fine. Matt From matt at ucc.asn.au Wed Mar 30 13:57:47 2005 From: matt at ucc.asn.au (Matt Johnston) Date: Wed Mar 30 13:58:00 2005 Subject: [tech] Morwong PSU Message-ID: <20050330055746.GA2653@morwong.ucc.gu.uwa.edu.au> Morwong turned off when I was sitting in the machineroom, it wouldn't turn back on. (Perhaps it would have if I'd pressed the correct button, being 'reset'. eh). The "hyena" brand ATX PSU was very hot, so I took it out. As expected the fan didn't spin freely. I've replaced it with a less dirty "seasonic" brand power supply from a box on the floor which had a sticker saying "bananabox". It seems to work now. Matt