On Wed, 14 Oct 1998 jaharkes_at_cs.cmu.edu wrote:

> > I'm trying to connect to testserver.coda.cs.cmu.edu as a client with
> > venus, and running into the following messages:
> > 
> > 20:27:00 GetRootVolume: can't get root volume name!
> > 20:27:15 GetRootVolume: can't get root volume name!
> > coda_psdev_write: downcall, no SB!
> > coda_get_rootfid: error -4
> > coda_read_super: coda_get_rootfid failed with -4
> > coda_psdev_write: downcall, no SB!
> > coda_psdev_write: downcall, no SB!
> > 
> > (those were pasted from dmesg and /usr/coda/etc/console)
> 
> I've checked the server, and it is just about doing nothing. I can't tell
> from here if you actually reached the server or not. Are you behind a
> firewall?
> 
> .... Wait a minute 2.1.125 kernel? Then you probably might need to get a
> more recent kernel package:
> 
>     ftp...:/pub/coda/src/linux-coda-development-4.7.2-2.1.123.tgz
> 
> something has changed that made this one incompatible with development
> kernels before 2.1.123. (Peter, was this related to the change in the
> inode_file_operations structure in the 11x kernels?)
> 
> let me know if that solves something for you.

When I initially tried the linux-coda-development and coda-4.7.2
distribution, I manually merged the updated coda vfs files into the linux
kernel tree.. in an attempt to compile coda support directly into the
kernel (not as a module)

That turned out to be the problem. I noticed theres #ifdef's in psdev.c
for example that outputs a more recent version number only when its used
as a module. I'll wait for the next patch to compile it as non-module.

I now have 2.1.125 with linux-coda-development-4.7.2 loaded as a module,
and the 4.7.2 client utils. The iname and getrootfid errors are gone and I
only have the SB and GetRootVolume messages coming up, which are supposed
to be normal..

Unfortunately /coda still shows NOT_REALLY_CODA and I've let venus sit for
a while.. its spewwed out hundreds of GetRootVolume messages already and
nothing new seems to be happening.. 

Here's venus.log from the venus.cache directory:

--
[ X(00) : 0000 : 21:37:12 ] Coda Venus, version 4.6 (28)
[ X(00) : 0000 : 21:37:12 ] LOGFILE initialized with LogLevel = 0 at Sat Oct 17 21:37:12 1998

[ X(00) : 0000 : 21:37:12 ] ProfInit(): profiling is broken; fix it!
[ X(00) : 0000 : 21:37:12 ] E StatsInit()
[ X(00) : 0000 : 21:37:12 ] L StatsInit()
[ X(00) : 0000 : 21:37:12 ] RecovDataSizes: Log = 0x283a78, Data = 0xa0e9e0
[ X(00) : 0000 : 21:37:12 ] Recov_AllocateVM: allocated 1000 bytes at 20000000
[ X(00) : 0000 : 21:37:12 ] Recov_AllocateVM: allocated a09000 bytes at 21000000
[ X(00) : 0000 : 21:37:15 ] Recov_DeallocateVM: deallocated 1000 bytes at 20000000
[ X(00) : 0000 : 21:37:15 ] Recov_DeallocateVM: deallocated a09000 bytes at 21000000
[ X(00) : 0000 : 21:37:36 ] E adviceserver::adviceserver: AdviceServer    

[ A(13) : 0000 : 21:37:36 ] adviceserver::main()

[ X(00) : 0000 : 21:37:36 ] NotifyUserOfServerDownEvent

[ H(07) : 0000 : 21:37:36 ] HDBDaemon about to sleep on hdbdaemon_sync
--

The ServerDownEvent looks suspicous.. comments?

Sorry if this is really basic stuff :)

- lv