Messages in this thread | | | Subject | Re: BUG REPORT: NFS/SMP file corruption | From | Trond Myklebust <> | Date | 02 Feb 2000 19:23:49 +0100 |
| |
>>>>> " " == Richard Ems <r.ems@gmx.net> writes:
> Are the possible file corruptions or differences with the > "soft" option supposed to be there? If so it's a very dangerous > option! While running my tests yesterday night I got a couple > of "nfs: server 'hostname' not responding" and some of the > corrupted files were at these times, but not all of them. I > was running the same tests as explained above (but with > "soft,timeo=30" and without the rsize and wsize options), and I > got 28 from 40 files corrupted. 13 of them where at the same > time as the server was not responding (why not??? no idea!) but > the other 15 became corrupted at other times!
Yes. I've heard a couple of the Sun engineers bemoan the 'soft' mount option and state that it should never have been invented...
Normally if the partition is soft mounted, an NFS operation is allowed to fail upon the first major timeout and will send the '-EIO' error back to the application. If you are using soft mounts, and are not checking the read/write errors, then that is indeed a bug in your program...
Cheers, Trond
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |