Time |
S |
Nick |
Message |
13:53 |
|
garth |
hey folks, I having an issue with the ze9.50 daemon on 1.2.3, due to work overload I'm not able to update koha to 2.x, I get the following eror |
13:54 |
|
garth |
Processing author="test" at 1 z3950.loc.gov:7090 voyager (1 forks) |
13:54 |
|
garth |
Q: @attr 1=1003 "test" |
13:54 |
|
garth |
ERROR: Can't call method "option" on an undefined value at /home/eigcampus/domains/library.eigcampus.org/public_html/intranet/cgi-bin/acqui.simple/z3950/processz3950queue line 163, <KC> line 20. |
13:55 |
|
garth |
any thoughts folks? I would like to at least have z39.50 running on the new server |
14:38 |
|
kados |
garth so the error is happening on koha 1.2.3 or koha 2.2? |
15:50 |
|
owen |
z3950 problems seem to always top the list of frequently-asked questions. |
16:00 |
|
kados |
yep ... but not server problems ;-) |
16:00 |
|
kados |
the client is much more complex |
16:01 |
|
owen |
Did Paul write the client? |
16:01 |
|
owen |
I'm assuming that was part of the original MARC work? |
16:01 |
|
kados |
I can't remember now ... I think chris may have done the original for 1.x |
16:01 |
|
owen |
Hm. 1.2.3 seems so long ago now. |
16:02 |
|
kados |
yea. |
16:05 |
|
owen |
kados, have you had time to work on any of your Koha special projects lately? |
16:06 |
|
garth |
hi folks, I don't have the time right now to update koha to 2.2 so I was just wondering if you guys had a quick fix for the 1.2.3 z39.50 problem I have. If not, the user will have to wait until I update to 2.x |
16:06 |
|
kados |
garth: I don't know of a quick fix |
16:06 |
|
kados |
garth: have you checked the mailing list archive? |
16:07 |
|
owen |
Hard for us to diagnose since we don't have a 1.2.3 installation |
16:08 |
|
kados |
there are two mailing list archives listed at |
16:08 |
|
kados |
http://www.koha.org/mailing/ |
16:08 |
|
kados |
you could search there for a similar error |
16:08 |
|
garth |
yeah, I was wondering, do you see alot of z39.50 errors in the 2.x series? |
16:09 |
|
kados |
owen: I haven't had a chance lately. The amazon and spellcheck stuff is almost ready to be committed into 2.3 |
16:09 |
|
kados |
garth: we don't use that module at NPL |
16:09 |
|
garth |
The reason I ask is other than the z39.50 issues Koha has worked excellent for us |
16:09 |
|
kados |
garth: but I do recall quite a bit of the mailing list questions having to do with Z39.50 |
16:10 |
|
kados |
garth: Your best bet would be to post a message to the list with details on the version of Koha, what you're trying to do, and the exact error messages ... paul will likely write you back within a day or so |
16:11 |
|
kados |
owen: network outage? |
16:11 |
|
garth |
ok, I was just wondering if anyone here has had experience with z39.50 in the 2.x series |
16:12 |
|
kados |
garth: well I wrote the Z-Server ... but that's different ;-) |
16:12 |
|
garth |
cool |
16:12 |
|
kados |
garth: and it requires MARC |
16:12 |
|
kados |
although ... it would be possible to do it without Koha's MARC |
16:13 |
|
garth |
My upgrade path is to move to 2.2 with MARC for my current 1.2.3 installtion |
16:13 |
|
garth |
And the client LOOOVES z39.50!! |
16:14 |
|
kados |
yea ... it's pretty important to our operation too ... we use BookWhere to nab records ... and we do resource sharing with our Z39.50 server |
16:14 |
|
kados |
anyone seen this link? http://www.libraryelf.com/ |
16:14 |
|
kados |
pretty neat service |
16:15 |
|
kados |
garth: I'd recommend investing your time in getting 2.2 up rather than getting Z39.50 working with 1.2.3 ... but that's just my two cents |
16:17 |
|
garth |
I agree completely |
16:38 |
|
kados |
si around? |
16:39 |
|
si |
indeed I am |
16:39 |
|
kados |
I just got my first soekris 4801 |
16:39 |
|
si |
ecellent |
16:39 |
|
kados |
any pointers? |
16:40 |
|
kados |
I'd like to run one of the BSDs |
16:40 |
|
kados |
and do firewall/transparent proxy/dhcp/dns cacheing |
16:41 |
|
kados |
I don't have a BDS installed on my laptop but I just made room on my HD to install it in case it's needed (looks like it from what I can tell) |
16:41 |
|
si |
mmm |
16:41 |
|
kados |
s/BDS/BSD/ ;-) |
16:41 |
|
si |
if you just want something that works out of the box with very little effort, then monowall is fun |
16:42 |
|
si |
but I'm not sure whether it'll manage transparent proxying |
16:42 |
|
kados |
do you do any of that at katipo? |
16:42 |
|
si |
any of which? |
16:43 |
|
kados |
transparent proxying |
16:43 |
|
si |
the BSD stuff, or transparent proxying? |
16:43 |
|
si |
I have a colleague of genuine Irish extraction |
16:43 |
|
kados |
:-) |
16:43 |
|
si |
who is fond of labelling various things "the devils work" |
16:43 |
|
kados |
even better :-) |
16:43 |
|
si |
NAT |
16:43 |
|
si |
Windows |
16:44 |
|
kados |
it's true ... transparent proxies fall into that category |
16:44 |
|
si |
and other things that I don't immediately recall |
16:44 |
|
kados |
:-) |
16:44 |
|
si |
and I suspect that he regards tp similarly |
16:44 |
|
kados |
well do you think m0n0BSD could handle it? |
16:44 |
|
si |
and I can't help but concur |
16:44 |
|
kados |
yep |
16:44 |
|
kados |
me too |
16:44 |
|
kados |
but the library wants it |
16:45 |
|
si |
do they? |
16:45 |
|
si |
or do they want proxying? |
16:45 |
|
kados |
for filtering |
16:45 |
|
si |
Neither my Irish friend nor I have any issues with proxying |
16:45 |
|
si |
it's monkeying with the TCP streams that we find objectionable |
16:45 |
|
kados |
well they want 'patron controlled filtering that doesn't require an initial login but only prompts for a login when they visit a questionable site and that's centralized and doesn't require browser muching' |
16:45 |
|
kados |
s/muching/mucking/ |
16:46 |
|
si |
both mozilla and IE have mechanisms to centrally configure them via Javascript |
16:46 |
|
si |
you can use that to force use of proxies |
16:46 |
|
kados |
from where do you configure them? |
16:47 |
|
si |
you point them at a carefully crafted page on your webserver |
16:47 |
|
kados |
hehe |
16:48 |
|
si |
"Automatic proxy configuratino URL" |
16:48 |
|
si |
it's in the firefox proxies tab |
16:48 |
|
kados |
hmmm, well that's an option |
16:48 |
|
kados |
right ... well the problem is that not all of our machines are locked down enough |
16:49 |
|
kados |
so patrons could potentially override the filter (i.e., teenage boys who know more about computers than my staff) |
16:49 |
|
kados |
a transparent proxy (we have one now that we pay lots for) eliminates the need |
16:49 |
|
si |
just firewall direct access off at the edge |
16:49 |
|
kados |
hmmm ... tell me more |
16:50 |
|
si |
however, if it's only members of the public using it, what do you care if the tranny proxy mucks them up :-) |
16:50 |
|
kados |
right :-) |
16:50 |
|
si |
well, if you set machines to use a standard proxy - squid, apache, or something commercial |
16:51 |
|
si |
then from the POV of your NAT firewall, all requests are coming from the cache, not from the end users |
16:51 |
|
si |
so you can firewall the endusers internally |
16:51 |
|
si |
and not allow them to make outbound http/https/ftp requests |
16:51 |
|
si |
we did this for the opacs at HLT, to force them to use the proxy |
16:51 |
|
si |
ie, if they disabled the proxy, nothing worked |
16:51 |
|
kados |
that sounds like a nice setup |
16:52 |
|
si |
it's pretty low rent, and does require that the machines be configured to use the proxy |
16:52 |
|
kados |
our network is pretty messy currently |
16:52 |
|
si |
but it does tend to make the browsing experience a little more reliable |
16:52 |
|
kados |
yea .. |
16:53 |
|
si |
than using transparent proxies |
16:53 |
|
si |
which are the devils work :-) |
16:53 |
|
kados |
:-) |
16:53 |
|
si |
anyhoo, choices over how you implement your proxy is kind of up to you, if you can cope with the vagueness that transparent proxying introduces, then run with it |
16:54 |
|
si |
how many users are you going to be putting through the proxy, and at what sort of bandwidth utilisations? |
16:55 |
|
si |
and what software do you propose to use for the tp? |
16:55 |
|
kados |
about 200 users at about 1.0Mb spread over seven locations |
16:55 |
|
kados |
maybe squid or squidguard |
16:56 |
|
kados |
our network is so messy currently it's hard to even talk about |
16:56 |
|
kados |
we used to be fairly centralized with one gateway router at NPL |
16:56 |
|
kados |
hmmm, we've been using IPrism which as far as I can tell is a tp |
16:56 |
|
kados |
but costs a pretty penny |
16:57 |
|
kados |
(for us) |
16:57 |
|
si |
I don't know if they've been resolved or not, I think at the time they looked pretty fundamental |
16:58 |
|
kados |
well I guess as long as I had a way to avoid having to change every browser, and could still offer temporary overrides and not force any kind of login I'd be ok with any solution |
16:58 |
|
kados |
thos are the three main requirements |
16:59 |
|
si |
yes, 200 desktops is a fair number |
16:59 |
|
kados |
I thought the easiest way to solve our current overall network prob would be to put a firewall at each branch that did transparent proxying and dhcp + dns cacheing |
16:59 |
|
si |
so what's wrong with the current box? |
16:59 |
|
kados |
we don't have a firewall currently at any location |
16:59 |
|
kados |
and our networks are decentralizing |
17:00 |
|
kados |
so we're losing our 'gateway' |
17:00 |
|
si |
righto, so your libraries are moving onto DSL-like services that are connected to the interweb directly? |
17:00 |
|
kados |
some have gone to DSL, some to Cable Modem |
17:00 |
|
kados |
yep |
17:01 |
|
kados |
so firewalling is really important now |
17:01 |
|
si |
and by their assymetrical nature, you don't want to haul that data back to a central point via VPNs |
17:01 |
|
kados |
(before our ISP sorta handled that ) |
17:01 |
|
kados |
right! |
17:01 |
|
si |
modems |
17:01 |
|
si |
and dsl |
17:01 |
|
si |
are more of my Irishmans devils work... |
17:01 |
|
kados |
:-) |
17:02 |
|
si |
he freely acknowledges that the devil is a busy entity :-) |
17:02 |
|
kados |
hehe |
17:03 |
|
si |
are you after the caching aspects of a proxy? |
17:03 |
|
si |
or just the content mediation? |
17:03 |
|
kados |
nope |
17:03 |
|
kados |
just content mediation |
17:03 |
|
kados |
dns caching would be nice for security reasons |
17:03 |
|
kados |
but that's all |
17:03 |
|
si |
DNS caching is a much less intensive task |
17:04 |
|
si |
you can cache a lot of DNS in 1MB of RAM |
17:04 |
|
si |
but sweet FA of the interweb |
17:04 |
|
kados |
well ... here's the other side of this coin ... we want to use our Koha database to determine who has override privileges :-) |
17:05 |
|
kados |
and I'd like to use the dmoz directory to determine what falls into the 'moderated' category (currently only porn and chat do) |
17:05 |
|
si |
you get 128MB RAM with the 4801? |
17:05 |
|
si |
dmoz? |
17:05 |
|
kados |
open directory = dmoz |
17:06 |
|
kados |
dmoz.org |
17:06 |
|
si |
ahh |
17:06 |
|
kados |
I think I can do a CF on the 4801 |
17:06 |
|
rach |
hello |
17:07 |
|
kados |
hey rach |
17:07 |
|
si |
dmoz.org provides some kind of filtered access? |
17:07 |
|
kados |
no ... just a human edited directory with nifty categories like 'porn' |
17:07 |
|
si |
so how do you use it at the moment? |
17:08 |
|
kados |
we don't use dmoz atm |
17:08 |
|
kados |
IPrism maintains a database of their own |
17:08 |
|
si |
so it does |
17:08 |
|
kados |
this would be the 'free' version |
17:09 |
|
si |
so you want the punters to have free access until they go somewhere dubious |
17:09 |
|
si |
and at that point you want them to auth |
17:09 |
|
kados |
yep |
17:10 |
|
si |
and you don't yet have a machine readable definition of dubious |
17:10 |
|
kados |
and the list of who can auth comes from our borrower table |
17:10 |
|
si |
that breaks down to a list of URL's? |
17:11 |
|
kados |
no ... but I'm pretty sure that's easy to get out of dmoz |
17:11 |
|
kados |
yep there is a CF slot here |
17:12 |
|
kados |
so that'd take care of storage issues I should think |
17:12 |
|
si |
there is active blacklisting in squidguard |
17:13 |
|
kados |
yep ... we'd like to have that list oo |
17:13 |
|
kados |
too even |
17:13 |
|
kados |
last I checked I don't think squidguard did transparent proxies |
17:14 |
|
si |
sg sits inside squid |
17:14 |
|
kados |
and since the last release was Dec 2001 I'd say it hasn't changed |
17:14 |
|
si |
so you presumably need squid to handle the actual heavy lifting |
17:15 |
|
kados |
probably ... though recently I was wondering if I could do the whole thing with iptables and linux |
17:15 |
|
kados |
and maybe some perl ;-) |
17:15 |
|
kados |
but that's probably a bad idea |
17:16 |
|
kados |
my tcp/ip isn't that strong |
17:16 |
|
si |
mm, ok |
17:16 |
|
si |
you've never setup a tranny proxy before? |
17:17 |
|
kados |
nope |
17:17 |
|
si |
the answer, in a nutshell, is that you need all those things |
17:17 |
|
kados |
hehe |
17:17 |
|
si |
you need some device that sees all the traffic, to pick up the http requests |
17:17 |
|
si |
normally that would be your NAT firewall |
17:18 |
|
si |
but it could be some other bridging device that you stick live in the network |
17:18 |
|
kados |
my idea was that the soekris would be the NAT firewall |
17:18 |
|
si |
either way, it has to see the entire communication between the client, and the interweb |
17:18 |
|
kados |
right |
17:19 |
|
si |
once you've got that, and iptables or pf or similar, then you can redirect the http/https traffic to another TCP port |
17:19 |
|
si |
that TCP port could be on the same machine, or it could be elsewhere on your network |
17:20 |
|
si |
there is no particular requirement that hte transparent proxy be on the same machine as the "http stealing" firewall rules |
17:20 |
|
kados |
right ... and by TCP port you mean a logical port ... could be the same interface right? |
17:20 |
|
si |
listening on that TCP port is a process that knows how to read the HTTP request, and service it |
17:20 |
|
si |
ie, squid, or similar |
17:21 |
|
kados |
ok |
17:21 |
|
si |
so at that point, squid is deciding what to do with the http request, based on it's configuratino, which may include lookups through squidguard |
17:22 |
|
kados |
(this is starting to make sense) ... |
17:22 |
|
kados |
I was wondering how the two interacted |
17:22 |
|
si |
it is certainly easiest to do all this on the same box, but you could in theory have your firewall redirects, your squid, and your squidguard all on seperate boxes |
17:23 |
|
kados |
right ... |
17:23 |
|
kados |
good news then |
17:23 |
|
si |
now, here in NZ, we normally run squid for content caching |
17:23 |
|
si |
since we're a stupidly long way from anywhere |
17:24 |
|
kados |
:-) |
17:24 |
|
si |
and trans pacific bandwidth costs amounts of money that truly would make your Ohio eyes water |
17:24 |
|
kados |
I'm sure |
17:24 |
|
si |
so there's real incentive to not use it, if you can avoid it |
17:25 |
|
si |
apparently Cisco's content caching divisino is actually housed in Australia, for much the same sorts of reasons |
17:25 |
|
si |
all the caching gurus are down under, and not in the US |
17:25 |
|
kados |
:-) |
17:25 |
|
si |
so we sould normally spec quite a big box for caching |
17:25 |
|
si |
since you'r aiming to store a pile of stuff |
17:26 |
|
si |
so you need much disk, and thus CPU and RAM as well |
17:26 |
|
kados |
right ... |
17:26 |
|
si |
so in rigs I've built down here, we've had the firewall running fairly lightweight, off flash, redirecting to a seperate box running squid |
17:27 |
|
si |
in terms of the 4801, you can either run with CF, or you can get a 2.5inch HD mount |
17:28 |
|
si |
what OS do you use for servers? |
17:28 |
|
kados |
currently linux (mix between debian and redhat/fedora) |
17:28 |
|
kados |
but I'd like to branch into the BSDs to expand my knowledge a bit (iykwim) |
17:28 |
|
si |
and you're thinking BSD for the 4801? |
17:29 |
|
kados |
yea |
17:29 |
|
si |
fair enough |
17:29 |
|
si |
for ease of setup and use, it's pretty hard to go past debian on CF |
17:29 |
|
kados |
I'm not opposed to linux tho if it'd do the job as well |
17:30 |
|
si |
we're using debian on 1GB flash |
17:30 |
|
si |
basic debian takes about 270MB |
17:30 |
|
si |
so that gives you a certain amount of room to play in |
17:30 |
|
kados |
right |
17:30 |
|
kados |
sounds promising |
17:30 |
|
si |
where you need to be careful is that you have enough RAM for everything you want to do |
17:30 |
|
kados |
128MB sound like enough? |
17:31 |
|
si |
because you don't have a lot of swap available |
17:31 |
|
kados |
ahh ... right |
17:31 |
|
si |
it's certainly more than enough for the basic firewalling and NAT and whatnot |
17:31 |
|
si |
since that all takes place in the kernel |
17:32 |
|
si |
you'll get change from 32MB if that's all you do with the box |
17:34 |
|
kados |
I'm thinking that the dmoz and squidguard lists aren't that big ... |
17:34 |
|
si |
byw way of comparison, this is a P4 box with 16 ethernet interfaces, that's been up for 65 days, and it runs ssh, snmpd, zebra, cron, inetd and a few gettys |
17:34 |
|
si |
it has a metric truckload of firewall rules |
17:34 |
|
si |
MemTotal: 256508 kB |
17:34 |
|
si |
MemFree: 225556 kB |
17:35 |
|
si |
it's using 31MB RAM, after 60 days of operation |
17:35 |
|
kados |
wow |
17:35 |
|
kados |
that's pretty cool |
17:36 |
|
si |
sadly, it crashes every four months or so, for reasons I haven't been able to fathom |
17:36 |
|
kados |
:-) |
17:37 |
|
si |
MemTotal: 63192 kB |
17:37 |
|
si |
MemFree: 43900 kB |
17:37 |
|
si |
19MB in use |
17:37 |
|
kados |
wow that's great |
17:38 |
|
si |
the difference is probably zebra, which can use an arbitrary amount of RAM, depending on how many routes there are in your route table |
17:38 |
|
kados |
I wonder how much hp squid would take up if it was just doing tp for a small list like dmoz and squidguard |
17:38 |
|
si |
from memory, squid uses loads of RAM, but uses it for caching objects, and keeping indexes of it's cache |
17:38 |
|
si |
if you're not caching, you'll probably be ok |
17:38 |
|
kados |
cool |
17:39 |
|
si |
I've no idea what the RAM utilisation of sg is, I've never run it |
17:39 |
|
kados |
right |
17:39 |
|
kados |
what debian distro do you use for your soekris? |
17:39 |
|
si |
we don't use debian on the soekris's |
17:39 |
|
kados |
ahh |
17:40 |
|
kados |
openbsd? |
17:40 |
|
si |
I've been experimenting with debian of flash on some more stock EPIA motherboards |
17:40 |
|
si |
which works ok, but the mobos in questino have at least 512MB RAM |
17:41 |
|
si |
because we're running apache, perl, mysql and so forth off the flash |
17:41 |
|
kados |
what's the application ? |
17:41 |
|
si |
there's absolutely no reason why a stock debian shouldn't work off flash in 128MB RAM, so long as you didn't go hog wild with extra clag |
17:41 |
|
si |
like X, and so forth |
17:42 |
|
si |
and kept the number of listening daemons to an absolute minimum |
17:42 |
|
kados |
right ... none of my debs have X (except one partition on my laptop) |
17:42 |
|
si |
my inclination would be to attach a hard disk with debian on it to your soekris |
17:43 |
|
si |
and turn off swap |
17:43 |
|
si |
and make sure that you can get everything you want to get running running |
17:43 |
|
si |
then transfer that to CF |
17:43 |
|
si |
you might want to invest in a CF <-> IDE adapter |
17:44 |
|
kados |
this soekris board came out of my pocket ... |
17:44 |
|
kados |
I didn't want to spend library money if it flopped :-) |
17:44 |
|
kados |
I've got a few CFs however |
17:46 |
|
si |
fair enough |
17:46 |
|
si |
any of the CF bigger than 256MB? |
17:46 |
|
kados |
256 is the biggest I've got atm |
17:46 |
|
si |
my experience is that getting debian in less than 256 is a bit tedious |
17:46 |
|
si |
to the point where you might as well start playing with the CF specific distros |
17:46 |
|
si |
on our soekris's we use Bering |
17:47 |
|
kados |
bering eh? |
17:47 |
|
si |
which takes up about 8MB of flash for our default load |
17:47 |
|
kados |
:-) |
17:47 |
|
si |
which in fact includes 1.2MB of DOS6.22 :-) |
17:47 |
|
kados |
:-) |
17:47 |
|
kados |
ahh Dos6.22 |
17:47 |
|
kados |
those were the days |
17:48 |
|
si |
yet another of the devils works |
17:48 |
|
kados |
dad deleting it |
17:48 |
|
kados |
and the wonders of undelete ;-) |
17:48 |
|
si |
and my irishman isn't even a catholic :-) |
17:48 |
|
kados |
:-) |
17:49 |
|
kados |
so would you recommend bering for squid/sg? |
17:49 |
|
si |
to save yourself going utterly mental, you might want to go for one of these: |
17:49 |
|
si |
http://www.pcengines.ch/cflash.htm |
17:49 |
|
si |
although finding one closer than switzerland would be the go |
17:50 |
|
kados |
wow ... that does look nice |
17:50 |
|
kados |
I could install the os on a regular pc and then just plug it in, eh? |
17:51 |
|
si |
yes |
17:51 |
|
kados |
nice! |
17:51 |
|
si |
it saves a whole bunch of grief around getting the thing bootable |
17:51 |
|
kados |
yea ... I was dreading that |
17:51 |
|
kados |
mucking about with serial and such :-) |
17:52 |
|
si |
which can be difficult to do in pcmcia or USB adapters |
17:52 |
|
si |
since getting a CDROM drive or similar onto the soekris is non trivial, it's easier to do it on a more standard machine |
17:53 |
|
si |
althougk, I do note that the bering people now have a pxelinux rig |
17:53 |
|
si |
which also may be worth a look |
17:53 |
|
si |
http://leaf.sourceforge.net/be[…]&MMN_position=3:3 |
17:53 |
|
si |
http://leaf.sourceforge.net/in[…]op=view&ANN_id=37 |
17:54 |
|
si |
things to watch out with the soekris, are that not all CF are created equal |
17:54 |
|
kados |
I've had the same problem with my Zaurus |
17:54 |
|
si |
you may struggle to get older CF bootable |
17:55 |
|
si |
sandisk CF have been good for us |
17:55 |
|
si |
there looks to be a squid package for bering |
17:56 |
|
si |
we use a homebuilt PXE installation to setup our soekris's |
17:56 |
|
kados |
I've got a lexar CF ... no sandisks |
17:56 |
|
si |
it might be the way to go for you |
17:56 |
|
kados |
yea? |
17:56 |
|
si |
you'll certainly learn some stuff :-) |
17:56 |
|
si |
PXE is netbooting |
17:56 |
|
si |
so you setup a DHCP server with some extra info, and a TFTP server |
17:57 |
|
si |
and the machines download their kernels from off the ethernet |
17:57 |
|
kados |
sounds a bit like LTSP |
17:57 |
|
si |
yes, a lot like |
17:57 |
|
kados |
yea I've got lots to learn |
17:57 |
|
si |
if you're familiar with ltsp, then setting up pxe to setup the soekris shouldn't be too hard |
17:58 |
|
kados |
well I've done ltsp ... I wouldn't say I'm particularly farmiliar with it ;-) |
18:01 |
|
si |
well, you've some kind of time<->pfaffing<->money tradeoff here |
18:01 |
|
si |
if you just wnat it to work, get a bigger CF and an IDE convertoer, wack it in an ordinary PC to do setup, and you'll be good to go licketysplit |
18:01 |
|
kados |
:-) |
18:01 |
|
si |
if you want to muck around and learn some for the leat amount of dollars spent |
18:02 |
|
si |
then stick a 32MB CF in, and muck around with PXE Installs |
18:02 |
|
si |
you'll have great fun |
18:02 |
|
kados |
hehe |
18:02 |
|
kados |
I think I'll do both while I wait for the IDE convertor to arrive :-) |
18:02 |
|
si |
monowall is definitely worth a look, if only because it's real easy to setup |
18:03 |
|
si |
I've not played with the other bsd based flash boot systems |
18:03 |
|
si |
they all looked like they weren't as well sorted as bering |
18:03 |
|
si |
which has been around a lot longer |
18:03 |
|
si |
the BSD people weren't as quick out of the blocks in getting their OS's running out of RAM disks |
18:03 |
|
si |
ohh, one thing to bear in mind |
18:04 |
|
si |
is that debian will be less consumptive of RAM than bering |
18:04 |
|
si |
Bering boots from your boot media, and extracts all the files into a ram disk |
18:04 |
|
si |
of 6-12MB |
18:04 |
|
si |
which you lose as operating RAM |
18:04 |
|
kados |
ahh |
18:04 |
|
si |
that's one of the tradeoffs |
18:05 |
|
kados |
are new kernels being used (2.6es?) |
18:05 |
|
si |
since Bering was and is still designed to run off a floppy |
18:05 |
|
si |
so the softwar on it is compressed to death |
18:05 |
|
si |
so you can't run from it directly |
18:05 |
|
kados |
ahh |
18:05 |
|
si |
where as with a stock debian install, you just run straight from CF |
18:06 |
|
kados |
makes sense ... it needs to eb decompressed before the files are readable ... |
18:06 |
|
kados |
gotcha |
18:06 |
|
si |
OTOH, the bering style systems won't wear out your CF, but the debian install will |
18:06 |
|
si |
although you could mitigate most of that by logging back to a central syslog server |
18:06 |
|
kados |
'wear out'? |
18:07 |
|
si |
urban myth suggests that CF have much lower duty cycles than spinning hard disks |
18:07 |
|
kados |
ahh |
18:07 |
|
kados |
well if I get a couple of years out of them I'll be happy |
18:07 |
|
si |
especially, that you only get a finite number of writes, which may not be very high |
18:08 |
|
si |
modern flash is apparently a lot better |
18:08 |
|
si |
flash is also somewhat slower than a spinning disk, somewhat counterintuitively |
18:08 |
|
si |
which is to say, it's seek times are much better, but it's transfer rates are dismal |
18:08 |
|
kados |
right I've heard that |
18:08 |
|
si |
so you get a somewhat more responsive system running the Bering way, out of RAMdisk |
18:08 |
|
kados |
that little disk spins pretty quickly for a reason ;-) |
18:09 |
|
si |
indeed it does |
18:09 |
|
si |
and once it's found the data, it hoovers it off pretty quickly |
18:09 |
|
kados |
hehe |
18:10 |
|
si |
as always, in the wonder world of open source, there are myriad ways to waste great wads of your time |
18:10 |
|
kados |
yep |
18:10 |
|
kados |
:-) |
18:10 |
|
si |
right now, however, it's a beautiful summers day here, the sun is pouring in the window |
18:10 |
|
si |
so I'm going to go help Rach train dogs |
18:10 |
|
kados |
sounds like fun |
18:10 |
|
kados |
thanks for the help si! |
18:11 |
|
si |
first day back after the summer break |
18:11 |
|
kados |
wow! |
18:11 |
|
si |
the pups will doubtless all be bigger |
18:11 |
|
si |
and awful :-) |
18:11 |
|
kados |
:-) |
18:11 |
|
si |
yes, you have to remember that here in NZ, Christmas and the summer break become one big holiday |
18:11 |
|
si |
since they fall at the same time of year |
18:12 |
|
si |
so there are lots of companies still closed until Monday |
18:12 |
|
si |
lots of people not yet back at work, and many won't be till the end of January |
18:13 |
|
si |
and things like recreational sport and clubs and whotnot tend to shutdown for a few weeks |
18:13 |
|
kados |
that's really nice |
18:13 |
|
kados |
things aren't as relaxed here |
18:13 |
|
si |
hence the reason that we've not done any dog training with the club since mid december |
18:13 |
|
kados |
I've thought about moving to NZ |
18:14 |
|
kados |
but haven't had a chance to visit yet |
18:14 |
|
si |
well, the downside is that we don't get a holiday in the middle of the year like the norther hemisphere does |
18:14 |
|
si |
since that's winter |
18:14 |
|
si |
but we do tend to bunk off and go snoeboarding then anyway |
18:14 |
|
kados |
:-) |
18:15 |
|
si |
hmm, wind is getting up http://www.huttcity.govt.nz/about/weather/ |
18:15 |
|
si |
could be a blokart day |
18:17 |
|
si |
later |
18:17 |
|
kados |
thanks again si |
18:18 |
|
kados |
wow that looks fun |
23:06 |
|
kados |
chris around? |
23:07 |
|
kados |
or si? |
02:03 |
|
kados |
si back from dog training? |
02:04 |
|
kados |
I've got my soekris booted with OpenBSD |
02:04 |
|
kados |
(what a chore PXE is!) |
02:04 |
|
kados |
but I think I understand the process now |
02:17 |
|
kados |
anyhoo ... if you happen by I've got it recognizing the 256MB CF card and I'm trying to sort out how to do the partitions |
02:17 |
|
kados |
(what the best allocation would be that is) |
02:53 |
|
kados |
hehe ... |
02:54 |
|
kados |
the good news is that open bsd is running on the soekris |
02:58 |
|
kados |
sweet it worked |
03:05 |
|
ambrose |
for your amusement, http://ada.dhs.org/~gniw/rcl.html pretty much summarizes the stupid parts of our issuing rules :-) |
03:08 |
|
kados |
ambrose: did you write that up? |
03:09 |
|
ambrose |
kados: yes, i wrote that up :-) |
03:09 |
|
kados |
:-) |
03:09 |
|
kados |
I agree with the last paragraph:-) |
03:10 |
|
ambrose |
:-) |
03:10 |
|
ambrose |
i think i'll get the videos catalogued with koha next week, whether my manager likes it or not. no one can search the video catalogue (which exists as just a simple printed list) right now |
03:11 |
|
ambrose |
i tried to do one video today and found that i don't know how to enter the call number |
03:20 |
|
ambrose |
indeed |