I have a program that watches the local utmp file and if theres a change
it copies it to /coda/system/hostname-utmp. So now all the other coda
clients on the network can run 'who' on these utmp files to see people
logged into other machines.
The problem as far I as can gather is that after some t
On Tue, 19 Sep 2000, Jan Harkes wrote:
> On Tue, Sep 19, 2000 at 12:47:19AM -0400, Christopher J Mason wrote:
> > Howdy.
> >
> > So call me crazy but I'm trying to get coda to work as a client on a
> > diskless machine (it happens to be an ARM based handheld called the iPAQ).
> > I've got ever
I'm now running kerberos5 version 1.2.1 and coda 5.3.8, clients and
servers all around.
I get the following in /vice/auth2/AuthLog whenever I run kclog anywhere:
08:16:51 In Krb5GetKeys()
08:16:51 Authentication failed for "nO0IAA" from x.x.x.x
The extra characters didn't show u
I've just recently upgraded from 5.3.5 (server and clients) on GNU/Linux
machines according to the release notes, and everything has gone
smoothly.
I'm still noticing occasional 'Connection timed out' errors on my diskless
coda client, with this new version. It happens after a day or two of
upti
On Mon, 17 Apr 2000, Chris Snyder wrote:
> Is anyone currently using Coda in a production environment? I am planning a
> web server cluster, and do not want to spend 1000's of dollars on a shared
> RAID system if I don't have to.
I would recommend trying coda for the web environment as you de
On Fri, 14 Apr 2000, Jan Harkes wrote:
> What's the problem you are seeing after a dirty shutdown? Maybe the fact
> that there is a problem in that case could be considered a Coda bug.
>
Oh, it turned out to be a stupid mistake on my part. I didn't look at the
error message carefully -- it was
Does anyone have a boot script that can gracefully run 'venus -init' at
boot time, if 'venus' on its own has trouble starting? This can happen
after the system hasn't been shut down normally, for example.
Can someone post a sample krb5.conf config file please?
On the server side, coda 5.3.5 kauth2 is balking on it. I run kclog on the
client and it says Krb5GetKeys, just as the kauth2 segfaults as it tries
to deal with /etc/krb5.conf. (discovered via an strace of kauth2)
I can kinit just fine on b
On Mon, 6 Mar 2000, Jan Harkes wrote:
> Also cfs/repair etc. will stop working as they talk to venus through a
> special control file in /coda. Is your cache stored in that ram-disk as
> well? Why not avoid the chroot and create some symlinks like:
>
> /usr -> /coda/usr
> /home -> /coda/home
>
I've tried running venus with '-rvmt 3' as was suggested on this list for
a diskless coda client machine, but it causes venus to die with an error
I've pasted below. This is with glibc and a Linux 2.2.14 kernel, coda
5.3.5.
I'm actually posting this from a diskless coda client, its working fine t
Quick question, do I need to recompile the coda server for kerberos
support to work, or just the client (venus) and generate kclog?
Also does anyone have a login or radius program which will inject kerberos
principles into the database, as a method of converting from /etc/shadow?
I checked the K
I'll be switching to kerberos next week so that users can be authenticated
automatically at login and have the right access permissions to the coda
directories, but I'm wondering in the meantime if theres a way to run clog
for each user on login so that they get access permissions, without having
Well I'm on the verge of testing a diskless coda system with 256mb ram. I
figure I can set aside a 20mb or 30mb ram drive for coda to use as local
cache -- for the time being I'll probably mount things as NFS though, i.e.
system programs and for mass storage purposes. But for hosting a
nameserver,
Here is a specific problem I'm having on a win98 machine with all the
latest windowsupdates. After some time the coda partition does get mounted
and I can view it graphically from the windows shell, read only, but
accessing it from a command prompt doesn't list any files. Maybe thats
normal for th
I'm not sure why but lately venus has been getting in a hung state every
now and then -- /coda cant be accessed. I'll try to peek through the logs
or turn on extra debugging to see whats up.
I've been running a 'stat' on /coda when a user logs in, in the hope that
I could have them change into their coda home directory only if it was
available. Which in turn is only done if their regular NFS home isn't
available.
The problem is I think when venus is in a messed up state, 'stat' does
Whats the command to make sure a file gets flushed back to the server?
Every so often it seems that when I reboot the machine running venus, it
doesn't come back up properly. I have to run venus -init to get it mounted
again. And to my surprise almost always an old version of the files I
was work
The tweaks needed were to redefine or change the 'link' and 'unlink'
functions in qmail's maildirdeliver and pine's maildir patch with 'rename'
and 'remove' respectively.
The only reason for using hardlinks for mail delivery, afaik, is to make
storage on nfs reliable. But since we're using coda,
Theres a program called 'maildirdeliver' at the qmail.org page, which
takes an email on input and delivers it into a directory with Maildir
format. A program like this can be set as default delivery on the system
or in a personal .qmail file in your home directory.
I was thinking this program cou
I checked the docs on cyrus and it stores messages individually in files,
as does qmail maildir, and I checked the source to confirm that it does in
fact use hardlinks extensivly..
Is there a clean way to allow hardlinks in coda between directories? As
far as I know, the reason links are used fo
For some reason I get the following error when making a hardlink to a file
in a different directory:
Cannot create hard link, "Invalid cross-device link"
Checking the venus source I see theres a verifyname call that makes sure
the file can't have any dots or slashes in it. This is in
vpr
I fetched the 3.2.4 win98 coda client distribution, and having uninstalled
my previous one, installed this one, and rebooted, codastart is able to
load venus but the whole kernel section (on the right) is faded out.
Any ideas why this might be? I'm running the latest win98, with ie 5.01
and all
This is kind of off-topic, but I can't think where else it might have been
brought up, or where I could have heard about it.
I'm looking for a project that is working to create a shared editting
environment, for editting lets say a text file in realtime, that would
update the changes with udp (o
On Tue, 29 Jun 1999, Juan Antonio Martinez wrote:
> On 28 Jun 1999, Soren Dayton wrote:
>
> ¿is coda the right way to any of these items? ¿has anyone did it before?
As long as you're not working with volumes approaching 20gig or 40gig or
so (at the moment), I think you should be ok.. coda isnt
I really like the maildir suggestion.
The one problem that I see is that if you have mail.example.com pointing
to both client coda machines for pop/imap work, a message might be
delivered to one of them before the other, in other words theres a race
condition.
The user contacts 'mail.example.co
On Sat, 17 Apr 1999, Roberto Diaz wrote:
> I'm trying to find out if is possible to use bootp over CODA in someway
> similar to NFS.
>
I would suggest getting the system booted as you probably have now over
NFS (root mounted NFS for example), and then just running 'venus' in your
startup script
As I understand things, one of the nice things about oracle is that the
servers can be set up with true redundency and fault tolerance. i.e.
multiple servers can fail and the database will still be accessible.
Would using gdbm or db (sleepycat db, included in GNU glibc), on a
distributed coda fil
On Tue, 19 Jan 1999, Troy Benjegerdes wrote:
> On Tue, 19 Jan 1999, Peter J. Braam wrote:
>
> > Where will the limits lie? I think that I can see that we can scale Coda
> > to approximately 500,000 files over the next year (the size of the files
> > is irrelevant). There will be implementation
The coda docs/howto dont explicitly state that an /etc/fstab entry isnt
needed.. is it? (This is on GNU/Linux)
Also just a suggestion, to have a Makefile option to compile just the
client or the server, much like there is an install option for them
independently. I imagine they both rely on the s
The following change is needed, at least for GNU/Linux running with the
more recent glibc's (currently alpha, around 2.0.110),
/* SFTP initialization. */
s = getservbyname("venus-se", 0);
That line needs to be changed to pass in either "udp" or "tcp" as the
second arguement, because 0 is
30 matches
Mail list logo
