jaffer-devel Mailing List for Jaffer
Status: Alpha
Brought to you by:
mrallen
Menu
▾
▴
jaffer-devel — Developer discussions
You can subscribe to this list here.
| 2003 |
Jan
|
Feb
(3) |
Mar
(19) |
Apr
|
May
|
Jun
|
Jul
(2) |
Aug
|
Sep
(3) |
Oct
|
Nov
|
Dec
|
|---|
|
From: Stewart A. <st...@ne...> - 2003-09-05 01:17:27
|
Yes, it does appear that the AFP server is in-kernel and does absolutely no sanity checking. I've nuked my dual G4 so many times it's not healthy. It's also really hard to troubleshoot this stuff. stewart Philip Edelbrock wrote: > > I think Apple implemented the client file-sharing protocol support in > the kernel(!?). It's very touchy and apparently a bit buggy. It > doesn't take much to confuse it and bring the machine down. :'( > > > Phil > http://www.baltra.org/ > > Maurice Volaski wrote: > >> I have jaffer compiled and running. However, it won't let me write >> files or folders to my mounted directory though it did accept my login. >> >> Also, I saw a kernel panic in OS X while I was mounted. How can jaffer >> be responsible for that? |
|
From: Philip E. <ph...@ed...> - 2003-09-04 22:02:34
|
I think Apple implemented the client file-sharing protocol support in the kernel(!?). It's very touchy and apparently a bit buggy. It doesn't take much to confuse it and bring the machine down. :'( Phil http://www.baltra.org/ Maurice Volaski wrote: > I have jaffer compiled and running. However, it won't let me write > files or folders to my mounted directory though it did accept my login. > > Also, I saw a kernel panic in OS X while I was mounted. How can jaffer > be responsible for that? |
|
From: Maurice V. <mvo...@ae...> - 2003-09-04 21:56:48
|
I have jaffer compiled and running. However, it won't let me write files or folders to my mounted directory though it did accept my login. Also, I saw a kernel panic in OS X while I was mounted. How can jaffer be responsible for that? -- Maurice Volaski, mvo...@ae... Computing Support, Rose F. Kennedy Center Albert Einstein College of Medicine of Yeshiva University |
|
From: Philip E. <ph...@ed...> - 2003-07-24 21:34:24
|
I looks like your path isn't complete. I.e., 'javac' isn't in your path. Check your Java installation and make sure the tools and libs are all accessable. Yeah, Netatalk still uses the original system that the U Mich guys arbitrarily picked a long time ago to store Mac resource forks on a Unix filing system. Hopefully they will eventually see the light and adopt a most standard OS-X way like what the Baltra patch does (see www.baltra.org if you haven't already). Phil Maurice Volaski wrote: > I am having trouble trying to compile under Gentoo Linux. > > I had to change in the makefile JDKDIR=/opt/blackdown-jdk-1.4.1, but > that wasn't enough. > > I also changed LIBDIR=${JDKDIR}/jre/lib/i386 and I am still getting a > complaint message... > > mkdir -p obj/lib > gcc -shared -o obj/lib/libLinux.so src/lib/OS_Server.c -Isrc/lib > -L/opt/blackdown-jdk-1.4.1/jre/lib/i386 -ljava -lcrypt > find src -name "*.java" | xargs javac -O -d obj > xargs: javac: No such file or directory > make: *** [javac] Error 127 > > > A couple of other questions: how does Jaffer stores double-fork files? > That is, does it use .AppleDouble directories? Also, how does it deal > with the DIDs? Netatalk doesn't appear to have been designed to handle > them very well. |
|
From: Maurice V. <mvo...@ae...> - 2003-07-24 21:21:01
|
I am having trouble trying to compile under Gentoo Linux.
I had to change in the makefile JDKDIR=/opt/blackdown-jdk-1.4.1, but
that wasn't enough.
I also changed LIBDIR=${JDKDIR}/jre/lib/i386 and I am still getting a
complaint message...
mkdir -p obj/lib
gcc -shared -o obj/lib/libLinux.so src/lib/OS_Server.c -Isrc/lib
-L/opt/blackdown-jdk-1.4.1/jre/lib/i386 -ljava -lcrypt
find src -name "*.java" | xargs javac -O -d obj
xargs: javac: No such file or directory
make: *** [javac] Error 127
A couple of other questions: how does Jaffer stores double-fork
files? That is, does it use .AppleDouble directories? Also, how does
it deal with the DIDs? Netatalk doesn't appear to have been designed
to handle them very well.
--
Maurice Volaski, mvo...@ae...
Computing Support, Rose F. Kennedy Center
Albert Einstein College of Medicine of Yeshiva University
|
|
From: Stewart A. <st...@ne...> - 2003-03-21 00:47:20
|
:-) sorry, Rick. it is quite early in development. while I have done some pretty serious stress tests (1000's files and 10's GB), it is also very possible to kernel panic your mac. the cvs version is probably a lot more reliable. stewart Rick Mann wrote: > So, finally tried to copy a file from the server to my desktop. No matter > what I try (permissions wise), the copy starts, I see the new icon appear, > then the Finder says I don't have permission, and the copy finishes and the > icon disappears. > > Furthermore, I've had two kernel panics (if you can call it that. you get > completely useless info now) and think I'm going to stop playing with Jaffer > for a while. > |
|
From: Rick M. <rm...@la...> - 2003-03-21 00:40:36
|
So, finally tried to copy a file from the server to my desktop. No matter what I try (permissions wise), the copy starts, I see the new icon appear, then the Finder says I don't have permission, and the copy finishes and the icon disappears. Furthermore, I've had two kernel panics (if you can call it that. you get completely useless info now) and think I'm going to stop playing with Jaffer for a while. -- Rick |
|
From: Rick M. <rm...@la...> - 2003-03-21 00:20:38
|
on 3/20/03 1:24 PM, Stewart Allen at st...@ne... wrote: > the native shared library probably didn't built correctly. if you look at > the jaffer.jar, does it contain lib/libLinux.so? where there any errors > reported on the server-side at runtime? No errors, just a "connect from" message from Jaffer. Also, Netatalk supports logging in using my First and Last names, rather than my Unix user name. Does Jaffer? -- Rick |
|
From: Rick M. <rm...@la...> - 2003-03-21 00:16:17
|
Is there anything more than the sample config file? I wanted to share a volume and turn off debugging, so I copied the sample and edited it a bit. I was able to log in, but then I got this additional password dialog after selecting a volume. When I tried to type my password back in, it failed to work. So I noticed the password = clever line, and commented it out, now I can't log in. -- Rick |
|
From: Stewart A. <st...@ne...> - 2003-03-20 23:22:03
|
sounds like you have debugging turned on in jaffer. turn it off if you want to see good performance for file transfers. it's not as much of a factor for directory listings. stewart Rick Mann wrote: >>that's an interesting (and cool) datapoint. I wonder what the difference is. > > > Okay, after you called it a datapoint, I decided I better actually measure > it once with some degree of repeatability. > > I have the same share under Netatalk 1.6.1 and > Jaffer-whatever-is-on-the-site. That directory on the server (RH 7.2, Java > 1.4.1_02) looks like this: > > www/ > com/ > latencyzero/ > 11 folders > 3 files (tar & tar.gz) > > Netatalk (afpd over tcp only) is listening on port 548, Jaffer on 5480. > Jaffer is running as root, in the background, and is spitting out data to a > terminal window. > > The connection is high/variable-latency, 256 kbit up, 2-3Mbit down. > > I mounted the Netatalk volume first, using the Finder. My Finder is set to > always open in a new window. The view is by icon. > > I double-clicked on the volume, then com. Then I wait for the menu bar clock > to hit xx:x5 seconds, and double-click on "latencyzero". By the time it > finishes fetching the icons for the files (the folders show immediately), 12 > seconds elapsed by the menu bar clock. > > I unmounted the Netatalk volume and mounted the Jaffer volume, and repeated > the process. It took 5 seconds (actually, less, because it caught me by > surprise) to completely fetch all the icons. > > I have not repeated this experiment, but it's consistent with my > (subjective) memory. > |
|
From: Rick M. <rm...@la...> - 2003-03-20 23:16:45
|
> that's an interesting (and cool) datapoint. I wonder what the difference is.
Okay, after you called it a datapoint, I decided I better actually measure
it once with some degree of repeatability.
I have the same share under Netatalk 1.6.1 and
Jaffer-whatever-is-on-the-site. That directory on the server (RH 7.2, Java
1.4.1_02) looks like this:
www/
com/
latencyzero/
11 folders
3 files (tar & tar.gz)
Netatalk (afpd over tcp only) is listening on port 548, Jaffer on 5480.
Jaffer is running as root, in the background, and is spitting out data to a
terminal window.
The connection is high/variable-latency, 256 kbit up, 2-3Mbit down.
I mounted the Netatalk volume first, using the Finder. My Finder is set to
always open in a new window. The view is by icon.
I double-clicked on the volume, then com. Then I wait for the menu bar clock
to hit xx:x5 seconds, and double-click on "latencyzero". By the time it
finishes fetching the icons for the files (the folders show immediately), 12
seconds elapsed by the menu bar clock.
I unmounted the Netatalk volume and mounted the Jaffer volume, and repeated
the process. It took 5 seconds (actually, less, because it caught me by
surprise) to completely fetch all the icons.
I have not repeated this experiment, but it's consistent with my
(subjective) memory.
--
Rick
|
|
From: Stewart A. <st...@ne...> - 2003-03-20 21:44:16
|
that's an interesting (and cool) datapoint. I wonder what the difference is. Rick Mann wrote: > I use Mac OS X connected to the 'Net via Sprint Broadband Direct, a > line-of-site terrestrial wireless internet access solution. > > The connection is very fast downstream, about 256 kbit upstream, and varying > but usually high latency. > > I connect to a linux box running Netatalk (and now Jaffer) in a Colo > facility across the Bay. It's a PIII 750, not much RAM (256 MB I think). It > has very little load. > > A Finder window that takes 10 seconds to populate with Netatalk populates in > 2 with Jaffer. Thank you. I'm eager to see what can happen with the New I/O > stuff... |
|
From: Stewart A. <st...@ne...> - 2003-03-20 21:28:49
|
the native shared library probably didn't built correctly. if you look at the jaffer.jar, does it contain lib/libLinux.so? where there any errors reported on the server-side at runtime? Rick Mann wrote: > Any ideas why my build of Jaffer wouldn't allow me to log in? I'd really > like to be able to make changes to Jaffer to see what niftiness I can > contribute, but if I can't build it, I can't contribute :-( > |
|
From: Rick M. <rm...@la...> - 2003-03-20 21:10:03
|
I use Mac OS X connected to the 'Net via Sprint Broadband Direct, a line-of-site terrestrial wireless internet access solution. The connection is very fast downstream, about 256 kbit upstream, and varying but usually high latency. I connect to a linux box running Netatalk (and now Jaffer) in a Colo facility across the Bay. It's a PIII 750, not much RAM (256 MB I think). It has very little load. A Finder window that takes 10 seconds to populate with Netatalk populates in 2 with Jaffer. Thank you. I'm eager to see what can happen with the New I/O stuff... -- Rick |
|
From: Rick M. <rm...@la...> - 2003-03-20 21:09:12
|
Any ideas why my build of Jaffer wouldn't allow me to log in? I'd really like to be able to make changes to Jaffer to see what niftiness I can contribute, but if I can't build it, I can't contribute :-( -- Rick |
|
From: Rick M. <rm...@la...> - 2003-03-20 21:04:18
|
on 3/20/03 12:29 PM, Stewart Allen at st...@ne... wrote: > error 2 is, I believe, an address error. did you misspell the server's name? > I know that sounds silly. also, make sure the mount point on the client > exists. I've had them disappear during testing before. So, with the combination of using your pre-built jar and remembering that I have to have a mount point (sorry 'bout that), I successfully logged in! WOW! Jaffer is AWESOMEly fast! Thank you very, very very much! Wow. I hope I'm able to use it reliably. I don't know what kinds of things are not yet implemented, but if I can put files on the server, copy them down, and delete them, that's all I really want. (Dunno what else there might be, but searches, moves on the server, etc, not so important). Thank you! -- Rick |
|
From: Stewart A. <st...@ne...> - 2003-03-20 20:33:36
|
error 2 is, I believe, an address error. did you misspell the server's name? I know that sounds silly. also, make sure the mount point on the client exists. I've had them disappear during testing before. Rick Mann wrote: > on 3/19/03 2:15 PM, Stewart Allen at st...@ne... wrote: > > >>oh, wait. i see that you built it yourself. perhaps there is a problem >>there. can you run the pre-built .jar file successfully? > > > Hmm. I dunno. > > Okay. I replaced my jaffer.jar with the one from the website, and now I get > > AFPMountURL returned error 2, errno is 2 > > |
|
From: Rick M. <rm...@la...> - 2003-03-19 23:46:41
|
on 3/19/03 2:15 PM, Stewart Allen at st...@ne... wrote: > oh, wait. i see that you built it yourself. perhaps there is a problem > there. can you run the pre-built .jar file successfully? Hmm. I dunno. Okay. I replaced my jaffer.jar with the one from the website, and now I get AFPMountURL returned error 2, errno is 2 -- Rick |
|
From: Stewart A. <st...@ne...> - 2003-03-19 22:19:10
|
oh, wait. i see that you built it yourself. perhaps there is a problem there. can you run the pre-built .jar file successfully? stewart Rick Mann wrote: > on 3/19/03 11:57 AM, Stewart Allen at st...@ne... wrote: > > >>I thought this was in the FAQ. Under any unix, you *have* to run as root to >>access the shadow passwords. Otherwise it wouldn't be 'secure' :-) > > > Yeah, I'm running it as root (see original email :-) > |
|
From: Rick M. <rm...@la...> - 2003-03-19 22:16:40
|
on 3/19/03 11:57 AM, Stewart Allen at st...@ne... wrote: > I thought this was in the FAQ. Under any unix, you *have* to run as root to > access the shadow passwords. Otherwise it wouldn't be 'secure' :-) Yeah, I'm running it as root (see original email :-) -- Rick |
|
From: Stewart A. <st...@ne...> - 2003-03-19 22:08:37
|
Rick, My reply to your email was caught in your spam filter. Check your inbox. In order to authenticate against shadow under unix, the process must be run as root. stewart Rick Mann wrote: > on 3/19/03 12:35 PM, Philip Edelbrock at ph...@ne... wrote: > > >>If the password is failing, check the logs (/var/log/secure and >>/var/log/messages). Perhaps you need to tweak the PAM config (I'm not >>sure exactly how Jaffer authenticates). > > > Hmm. That's not working. > > Heh, let me be more clear. mount_afp connects, but returns error -5023 > (Finder also complains that user/pass not valid). > > Neither messages nor secure show any activity when I attempt to log in. > |
|
From: Rick M. <rm...@la...> - 2003-03-19 21:07:42
|
on 3/19/03 12:35 PM, Philip Edelbrock at ph...@ne... wrote: > If the password is failing, check the logs (/var/log/secure and > /var/log/messages). Perhaps you need to tweak the PAM config (I'm not > sure exactly how Jaffer authenticates). Hmm. That's not working. Heh, let me be more clear. mount_afp connects, but returns error -5023 (Finder also complains that user/pass not valid). Neither messages nor secure show any activity when I attempt to log in. -- Rick |
|
From: Philip E. <ph...@ne...> - 2003-03-19 20:35:53
|
Do you mean that your username/password is failing, or that your client can't see the server at all? If your client can't see the server, then you could try to run the server on the usual port (is it 548?) and try to connect again or make sure you specify the port to OS-X when you try to connect (I think the syntax is something like afp://servername:portnumber/share ?). If the password is failing, check the logs (/var/log/secure and /var/log/messages). Perhaps you need to tweak the PAM config (I'm not sure exactly how Jaffer authenticates). Good luck! Phil Rick Mann wrote: >Hi. Jaffer is awesome! I really hope it gets reliable (if basic) soon! > >I just got the top of tree and built Jaffer on Linux RH 7.2 with Java >1.4.1_02 installed. I had to make some mods to the Makefile so that it would >find the libraries and jni headers, but after that it seemed to build fine. > >I've launched jaffer as root on port 5480, but I am unable to log in. Any >ideas what I should look for to determine what's failing? I believe my >system uses shadow passwords (I'm a Mac guy, not Unix). > >TIA! > > > |
|
From: Rick M. <rm...@la...> - 2003-03-19 19:56:23
|
Hi. Jaffer is awesome! I really hope it gets reliable (if basic) soon! I just got the top of tree and built Jaffer on Linux RH 7.2 with Java 1.4.1_02 installed. I had to make some mods to the Makefile so that it would find the libraries and jni headers, but after that it seemed to build fine. I've launched jaffer as root on port 5480, but I am unable to log in. Any ideas what I should look for to determine what's failing? I believe my system uses shadow passwords (I'm a Mac guy, not Unix). TIA! -- Rick |
|
From: Stewart A. <st...@ne...> - 2003-02-21 22:18:37
|
AFAIK, you're the first to try. If you get it working, I'm more than happy to roll in the changes. While this is a simple change, it's easier for me if you can provide context diffs. Is 'SunOS' the string returned for the system property 'os.name'?? This is used by the library loader (with spaces stripped). For example, OS X returns 'Mac OS X' which is turned into 'MacOSX' and then loaded as 'libMacOSX.so" stewart Aslak Raanes wrote: > Hello, > > Has anyone managed to build the jaffer C-library for Solaris? > > I changed Makefile and OS_Server.c to: > ... > (I hope it is ok to send the file. The diff would be become nearly the > same size anyway) > > # /store/gnu/bin/make lib-SunOS > mkdir -p obj/lib > gcc -shared -o obj/lib/libSunOS.so src/lib/OS_Server.c -Isrc/lib \ > -I/usr/java/include \ > -I/usr/java/include\solaris \ > gcc: : No such file or directory > In file included from src/lib/OS_Server.h:2, > from src/lib/OS_Server.c:1: > /usr/java/include/jni.h:27: jni_md.h: No such file or directory > In file included from src/lib/OS_Server.c:12: > /usr/include/shadow.h:18: warning: `SHADOW' redefined > src/lib/OS_Server.c:11: warning: this is the location of the previous > definition > make: *** [lib-SunOS] Error 1 > |
