Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Share FS
21 views
Skip to first unread message
Dave Plowman (News)
Nov 9, 2012, 6:08:50 AM11/9/12
to
I have two RPCs, (let's call them A & B) on the same LAN, and use ShareFS
between them. 'A' has a scanner, 'B' not, so I quite often have to swap
quite large files, usually from A to B. If I transfer from A to B at A it
starts doing it then says the HD can't be found, and B crashes with the HD
light left on, requiring a reset. If initiate the transfer at B, but still
A to B, it always works just fine.
A has a Unipod, B a 'normal' network card and a Blitz - both using IDEFS.
Both running RO 4.39.
--
*My dog can lick anyone
Dave Plowman da...@davenoise.co.uk London SW
To e-mail, change noise into sound.
between them. 'A' has a scanner, 'B' not, so I quite often have to swap
quite large files, usually from A to B. If I transfer from A to B at A it
starts doing it then says the HD can't be found, and B crashes with the HD
light left on, requiring a reset. If initiate the transfer at B, but still
A to B, it always works just fine.
A has a Unipod, B a 'normal' network card and a Blitz - both using IDEFS.
Both running RO 4.39.
--
*My dog can lick anyone
Dave Plowman da...@davenoise.co.uk London SW
To e-mail, change noise into sound.
Peter Young
Nov 9, 2012, 6:56:05 AM11/9/12
to
On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
> I have two RPCs, (let's call them A & B) on the same LAN, and use ShareFS
> between them. 'A' has a scanner, 'B' not, so I quite often have to swap
> quite large files, usually from A to B. If I transfer from A to B at A it
> starts doing it then says the HD can't be found, and B crashes with the HD
> light left on, requiring a reset. If initiate the transfer at B, but still
> A to B, it always works just fine.
> A has a Unipod, B a 'normal' network card and a Blitz - both using IDEFS.
> Both running RO 4.39.
This reply is from someone with near-total ignorance, so may be a red
> I have two RPCs, (let's call them A & B) on the same LAN, and use ShareFS
> between them. 'A' has a scanner, 'B' not, so I quite often have to swap
> quite large files, usually from A to B. If I transfer from A to B at A it
> starts doing it then says the HD can't be found, and B crashes with the HD
> light left on, requiring a reset. If initiate the transfer at B, but still
> A to B, it always works just fine.
> A has a Unipod, B a 'normal' network card and a Blitz - both using IDEFS.
> Both running RO 4.39.
herring. It could be a timing problem that ShareFS has. If so, it
might be cured by putting into Boof.Choices.Boot Tasks of both
machines an Obey file containing the line:
ShareFSWindow 1 { > null: }
Maybe someone will come along to explain that, but it solved a similar
problem for me.
With best wishes,
Peter.
--
Peter \ / \ Prestbury, Cheltenham, Glos. GL52
and \/ __ __ \ England.
family / / \ | | |\ | / _ \ http://pnyoung.orpheusweb.co.uk
/ \__/ \_/ | \| \__/ \______________ pny...@ormail.co.uk
Dave Plowman (News)
Nov 9, 2012, 9:39:26 AM11/9/12
to
In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
--
*Taxation WITH representation ain't much fun, either.
Peter Young <pny...@ormail.co.uk> wrote:
> This reply is from someone with near-total ignorance, so may be a red
> herring. It could be a timing problem that ShareFS has. If so, it
> might be cured by putting into Boof.Choices.Boot Tasks of both
> machines an Obey file containing the line:
> ShareFSWindow 1 { > null: }
Works a treat, thanks, Peter.
> This reply is from someone with near-total ignorance, so may be a red
> herring. It could be a timing problem that ShareFS has. If so, it
> might be cured by putting into Boof.Choices.Boot Tasks of both
> machines an Obey file containing the line:
> ShareFSWindow 1 { > null: }
--
*Taxation WITH representation ain't much fun, either.
Bryn Evans
Nov 9, 2012, 10:46:31 AM11/9/12
to
In a mad moment - Dave Plowman (News) mumbled :
> I have two RPCs, (let's call them A & B) on the same LAN, and use ShareFS
> between them. 'A' has a scanner, 'B' not, so I quite often have to swap
> quite large files, usually from A to B. If I transfer from A to B at A it
> starts doing it then says the HD can't be found, and B crashes with the HD
> light left on, requiring a reset. If initiate the transfer at B, but still
> A to B, it always works just fine.
> A has a Unipod, B a 'normal' network card and a Blitz - both using IDEFS.
> Both running RO 4.39.
Because of slight timing differences between machines it is always
better to Call a Large file from the receiving machine.
This seems to avoid a problem that I personally think, can be caused
by overrunning the receiving machines buffer.
--
|)����[
|)ryn [vans mail to - Bryn...@bryork.freeuk.com
> I have two RPCs, (let's call them A & B) on the same LAN, and use ShareFS
> between them. 'A' has a scanner, 'B' not, so I quite often have to swap
> quite large files, usually from A to B. If I transfer from A to B at A it
> starts doing it then says the HD can't be found, and B crashes with the HD
> light left on, requiring a reset. If initiate the transfer at B, but still
> A to B, it always works just fine.
> A has a Unipod, B a 'normal' network card and a Blitz - both using IDEFS.
> Both running RO 4.39.
better to Call a Large file from the receiving machine.
This seems to avoid a problem that I personally think, can be caused
by overrunning the receiving machines buffer.
--
|)����[
|)ryn [vans mail to - Bryn...@bryork.freeuk.com
Peter Young
Nov 9, 2012, 10:43:46 AM11/9/12
to
> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
> Peter Young <pny...@ormail.co.uk> wrote:
>> This reply is from someone with near-total ignorance, so may be a red
>> herring. It could be a timing problem that ShareFS has. If so, it
>> might be cured by putting into Boof.Choices.Boot Tasks of both
>> machines an Obey file containing the line:
>> ShareFSWindow 1 { > null: }
> Works a treat, thanks, Peter.
Good. Now will someone explain why? :-) Perhaps they have done, but
> Peter Young <pny...@ormail.co.uk> wrote:
>> This reply is from someone with near-total ignorance, so may be a red
>> herring. It could be a timing problem that ShareFS has. If so, it
>> might be cured by putting into Boof.Choices.Boot Tasks of both
>> machines an Obey file containing the line:
>> ShareFSWindow 1 { > null: }
> Works a treat, thanks, Peter.
if so I've forgotten.
Doug Webb
Nov 9, 2012, 11:39:58 AM11/9/12
to
In message <d8bffbeb5...@pnyoung.ormail.co.uk>
"A window in networking terms is how many packets that can be sent
before waiting for an acknowledgement.
With a window of 1, the sender must wait for each packet to be
acknowledged before it can send the next, so the transfer rate will be
low on connections with high latency.
On the other hand a large window will allow many packets to be
transmitted without waiting for the first acknowledgement, which will
be much faster and less dependent on the round trip latency. There is
a disadvantage on unreliable connections as it will take longer to
detect and recover from lost or corrupted packets.
With ShareFS problems it can sometimes help using a window of 1 to
slow down the transfer rate. Particular when a faster machine is
sending to a slower machine, who's network buffers can quickly become
exhausted."
Also ShareFS if I remember is using UDP protocols and it, UDP,
provides a minimal, unreliable, best-effort, message-passing transport
to applications and upper-layer protocols.
Doug
--
See and experience the future using ARM Technology - BeagleBoard -xM,
Cortex A8 and RISC OS 5.19.
Peter Young <pny...@ormail.co.uk> wrote:
> On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
>> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
>> Peter Young <pny...@ormail.co.uk> wrote:
>>> This reply is from someone with near-total ignorance, so may be a red
>>> herring. It could be a timing problem that ShareFS has. If so, it
>>> might be cured by putting into Boof.Choices.Boot Tasks of both
>>> machines an Obey file containing the line:
>>> ShareFSWindow 1 { > null: }
>> Works a treat, thanks, Peter.
> Good. Now will someone explain why? :-) Perhaps they have done, but
> if so I've forgotten.
> With best wishes,
> Peter.
Well as explained by Druck in 2011
> On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
>> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
>> Peter Young <pny...@ormail.co.uk> wrote:
>>> This reply is from someone with near-total ignorance, so may be a red
>>> herring. It could be a timing problem that ShareFS has. If so, it
>>> might be cured by putting into Boof.Choices.Boot Tasks of both
>>> machines an Obey file containing the line:
>>> ShareFSWindow 1 { > null: }
>> Works a treat, thanks, Peter.
> Good. Now will someone explain why? :-) Perhaps they have done, but
> if so I've forgotten.
> With best wishes,
> Peter.
"A window in networking terms is how many packets that can be sent
before waiting for an acknowledgement.
With a window of 1, the sender must wait for each packet to be
acknowledged before it can send the next, so the transfer rate will be
low on connections with high latency.
On the other hand a large window will allow many packets to be
transmitted without waiting for the first acknowledgement, which will
be much faster and less dependent on the round trip latency. There is
a disadvantage on unreliable connections as it will take longer to
detect and recover from lost or corrupted packets.
With ShareFS problems it can sometimes help using a window of 1 to
slow down the transfer rate. Particular when a faster machine is
sending to a slower machine, who's network buffers can quickly become
exhausted."
Also ShareFS if I remember is using UDP protocols and it, UDP,
provides a minimal, unreliable, best-effort, message-passing transport
to applications and upper-layer protocols.
Doug
--
See and experience the future using ARM Technology - BeagleBoard -xM,
Cortex A8 and RISC OS 5.19.
Peter Young
Nov 9, 2012, 1:10:12 PM11/9/12
to
On 9 Nov 2012 Doug Webb <doug....@btinternet.com> wrote:
> In message <d8bffbeb5...@pnyoung.ormail.co.uk>
> Peter Young <pny...@ormail.co.uk> wrote:
>> On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
>>> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
>>> Peter Young <pny...@ormail.co.uk> wrote:
>>>> This reply is from someone with near-total ignorance, so may be a red
>>>> herring. It could be a timing problem that ShareFS has. If so, it
>>>> might be cured by putting into Boof.Choices.Boot Tasks of both
>>>> machines an Obey file containing the line:
>>>> ShareFSWindow 1 { > null: }
>>> Works a treat, thanks, Peter.
>> Good. Now will someone explain why? :-) Perhaps they have done, but
>> if so I've forgotten.
>> With best wishes,
>> Peter.
> Well as explained by Druck in 2011
[snip useful explanation]
> In message <d8bffbeb5...@pnyoung.ormail.co.uk>
> Peter Young <pny...@ormail.co.uk> wrote:
>> On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
>>> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
>>> Peter Young <pny...@ormail.co.uk> wrote:
>>>> This reply is from someone with near-total ignorance, so may be a red
>>>> herring. It could be a timing problem that ShareFS has. If so, it
>>>> might be cured by putting into Boof.Choices.Boot Tasks of both
>>>> machines an Obey file containing the line:
>>>> ShareFSWindow 1 { > null: }
>>> Works a treat, thanks, Peter.
>> Good. Now will someone explain why? :-) Perhaps they have done, but
>> if so I've forgotten.
>> With best wishes,
>> Peter.
> Well as explained by Druck in 2011
Many thanks, and I've now locked the message, so I won't be able to
forget the explanation this time.
Bill (Adopt)
Nov 12, 2012, 8:24:24 AM11/12/12
to
In article <aa2709ec5...@pnyoung.ormail.co.uk>,
stored the message ..(safekeeping, of course)..! ;p))
Bill ZFC :))
--
Adoption InterLink UK with -=- http://www.billsimpson.com/
Domain Host Orpheus Internet -=- http://www.orpheusinternet.co.uk/
Peter Young <pny...@ormail.co.uk> wrote:
> On 9 Nov 2012 Doug Webb <doug....@btinternet.com> wrote:
> > In message <d8bffbeb5...@pnyoung.ormail.co.uk>
> > Peter Young <pny...@ormail.co.uk> wrote:
> >> On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
> >>> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
> >>> Peter Young <pny...@ormail.co.uk> wrote:
> >>>> This reply is from someone with near-total ignorance, so may be a red
> >>>> herring. It could be a timing problem that ShareFS has. If so, it
> >>>> might be cured by putting into Boof.Choices.Boot Tasks of both
> >>>> machines an Obey file containing the line:
> >>>> ShareFSWindow 1 { > null: }
> >>> Works a treat, thanks, Peter.
> >> Good. Now will someone explain why? :-) Perhaps they have done, but
> >> if so I've forgotten.
> >> With best wishes,
> >> Peter.
> > Well as explained by Druck in 2011
> [snip useful explanation]
> Many thanks, and I've now locked the message, so I won't be able to
> forget the explanation this time.
Now, all that remains is to remember where we've
> On 9 Nov 2012 Doug Webb <doug....@btinternet.com> wrote:
> > In message <d8bffbeb5...@pnyoung.ormail.co.uk>
> > Peter Young <pny...@ormail.co.uk> wrote:
> >> On 9 Nov 2012 "Dave Plowman (News)" <da...@davenoise.co.uk> wrote:
> >>> In article <93e7e6eb5...@pnyoung.ormail.co.uk>,
> >>> Peter Young <pny...@ormail.co.uk> wrote:
> >>>> This reply is from someone with near-total ignorance, so may be a red
> >>>> herring. It could be a timing problem that ShareFS has. If so, it
> >>>> might be cured by putting into Boof.Choices.Boot Tasks of both
> >>>> machines an Obey file containing the line:
> >>>> ShareFSWindow 1 { > null: }
> >>> Works a treat, thanks, Peter.
> >> Good. Now will someone explain why? :-) Perhaps they have done, but
> >> if so I've forgotten.
> >> With best wishes,
> >> Peter.
> > Well as explained by Druck in 2011
> [snip useful explanation]
> Many thanks, and I've now locked the message, so I won't be able to
> forget the explanation this time.
stored the message ..(safekeeping, of course)..! ;p))
Bill ZFC :))
--
Adoption InterLink UK with -=- http://www.billsimpson.com/
Domain Host Orpheus Internet -=- http://www.orpheusinternet.co.uk/
0 new messages