[Infrastructures] Coda, to do or not to do?

[Steve Wray]

Ivan Popov wrote:
> Hello Steve,
> 
> On Thu, Mar 17, 2005 at 01:32:42PM +1300, Steve Wray wrote:
> 
>>It turned out that if the client tries to access a file which is larger 
>>than the client side cache, the server would crash. Badly.
> 
> 
> are you sure you recall it correctly?

Yes; I had a diskless client with a coda cache on a 500M ramdisk.

Everything was great until I tried to play a 700M avi file on the client 
at which point the server would (very reliably) crash.

Playing a 400M file was no problem.

>>When I raised this on the coda mailing list (its probably in the 
>>archives...) it appeared that this was a design 'feature' and the 
[snip]
> 
> It would clean away a possible confusion, if you supply a reference
> to the relevant thread.

ahh my apologies; I just searched the archives for that thread and found 
another where I posted, but it was on a slightly different topic. I'm 
thinking now that it this client-cache-filling-crashing-server came up 
in private emails with Jan Harkes (the lead developer at the time IIRC).

I'll see if I can fish any of that out of my personal email archive.

Anyhow, its *easy* to test
:)
just be prepared for the server going down...


>>This was a few years ago. Perhaps its improved? But I certainly wouldn't 
>>trust data to it.
> 
> I appreciate your interest to Coda, even if you did not like it then.
> 
> On the other side, I have been on the Coda list for many years
> and I feel that the way you tell about your experiences may be misleading,
> especially a long time later.

it just seems that these days AFS is a more mature solution to the same 
problem...