Coda File System

Re: more observations

From: Peter J. Braam <braam_at_cs.cmu.edu>
Date: Fri, 18 Jul 1997 15:50:36 -0400 (EDT)
Bob,

I'll try to give you a few answers. 

Coda is always caching.  The files you fetch are stored in
/usr/coda/venus.cache.

coda-cache 1 switches on the attribute cache in the kernel (which is on by
default).  You can see what the kernel caches in /proc/net/coda-cache. 

The hangs should time out after one minute.  This is the time alloted for
a server to return to life after which Coda switchtes into disconnected
mode. 

The 0 inode numbers will slowly disappear. You don't need to worry too
much about them I think. 

Peter

On Mon, 14 Jul 1997 thoth_at_purplefrog.com wrote:

> 
>   OK, so I decide to really push the caching.  I activate the cache
> (coda-cache 1) and copy a 7M image into a directory in
> /coda/playground/thoth/.  the ls -l in another window shows the file
> steadily growing in size.  At some point this shows up in the codacon:
> 
> Store b16.tif [7729] ( 14:06:58 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (12) --> 49130 B/s ( 14:07:03 )
> Bandwidth jepthe.odyssey.cs..cmu.edu (6) --> 55268 B/s ( 14:07:13 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (13) --> 47645 B/s ( 14:07:18 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (9) --> 45964 B/s ( 14:07:23 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (8) --> 42313 B/s ( 14:07:28 )
> BeginRvmFlush (1, 14116, T) ( 14:07:33 )
> EndRvmFlush ( 14:07:33 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (14) --> 55944 B/s ( 14:07:33 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (11) --> 57628 B/s ( 14:07:40 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (13) --> 48385 B/s ( 14:07:45 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (15) --> 57231 B/s ( 14:07:50 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (17) --> 56963 B/s ( 14:07:55 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (15) --> 63430 B/s ( 14:08:00 )
> BeginRvmFlush (1, 1284, T) ( 14:08:05 )
> EndRvmFlush ( 14:08:05 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (16) --> 54992 B/s ( 14:08:05 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (18) --> 60594 B/s ( 14:08:10 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (10) --> 49223 B/s ( 14:08:15 )
> Bandwidth jepthe.odyssey.cs.cmu.edu (10) --> 47852 B/s ( 14:08:20 )
> 
>   This is accompanied by messages in the venus.log talking about
> "connection is strong" or "weak".  My guess is the threshold is 50KB/sec,
> which happens to be about the speed of my connection :)
> 
>   I do another ls -l, and it's hung.  We'll see if it ever gets un-hung. 
> The cp is still not complete.
> 
>   Slightly disappointing.  I consider it somewhat important that the
> filesystem should keep working, even if some of the reads/writes block.
> 
>  frop:21 $ cat /proc/meminfo
>         total:    used:    free:  shared: buffers:  cached:
> Mem:  64765952 63868928   897024 39297024  3735552 24240128
> Swap: 135761920        0 135761920
> 
> 
>   I wonder if the following messages from dmesg are at all alarming:
> 
> Coda Kernel/User communications module 0.04
> coda_read_super: rootfid is (-520093689, 1, 1)
> coda_fetch_inode: inode called with i_ino = 0
> coda_read_super: rootinode is 0 dev 5
> coda_fetch_inode: inode called with i_ino = 0
> 
> 
>   Well, now the ls is unhung and cp is complete.  At least it didn't
> crash.  The whole episode took about 9 minutes I think.
> 
> -- 
> Bob Forsman                                   thoth_at_gainesville.fl.us
>            http://www.gainesville.fl.us/~thoth/
> 
Received on 1997-07-18 16:16:17