* Re: autofs4: How to confg TCP 1st choice, not UPD?
[not found] <mailman.122.1162331942.2205.autofs@linux.kernel.org>
@ 2006-11-01 22:48 ` Risto Bell
2006-11-01 23:14 ` Jeff Moyer
0 siblings, 1 reply; 2+ messages in thread
From: Risto Bell @ 2006-11-01 22:48 UTC (permalink / raw)
To: autofs
Using RedHat Enterprise 3 update 6. Release notes consistent with
my experience that UDP is the default. Perhaps RedHat changed the
default autofs4's natural TCP back to UDP? In any event, question
remains: There isn't an simple config by which to control it (prefer
TCP with timeo=600,retrans=2 1st), it has to be compiled into it?
Also could use multi-mounts? List two alternative mounts, same but
one TCP and one UPD, so the 1st (TCP) will be attempted, and only if
it fails (server doesn't support TCP) will it fall to 2nd (UDP)?
This would still need programs/scripts but would save rcpinfo process,
and achieve slight speed-up.
% egrep -C4 ' autofs-4|NFS.*UDP' RELEASE-NOTES-en RELEASE-NOTES-U6-en
RELEASE-NOTES-en- o The Red Hat Enterprise Linux 3 kernel now includes support for NFS
RELEASE-NOTES-en- over TCP. To use NFS over TCP, you must include the "-o tcp" option to
RELEASE-NOTES-en- mount when mounting the NFS-exported file system on the client system.
RELEASE-NOTES-en-
RELEASE-NOTES-en: NOTE: The default transport protocol for NFS remains UDP. Use the
RELEASE-NOTES-en- mount command with the "-o tcp" option to mount an NFS-exported file
RELEASE-NOTES-en- system using TCP; otherwise, UDP will be used by default.
RELEASE-NOTES-en-
RELEASE-NOTES-en- o In this kernel, the following command has been added to scan for new
--
RELEASE-NOTES-U6-en- o anaconda-9.1.5.8-1.RHEL => anaconda-9.1.6.8-1.RHEL
RELEASE-NOTES-U6-en-
RELEASE-NOTES-U6-en- o anaconda-runtime-9.1.5.8-1.RHEL => anaconda-runtime-9.1.6.8-1.RHEL
RELEASE-NOTES-U6-en-
RELEASE-NOTES-U6-en: o autofs-4.1.3-130 => autofs-4.1.3-150
RELEASE-NOTES-U6-en-
RELEASE-NOTES-U6-en- o binutils-2.14.90.0.4-37 => binutils-2.14.90.0.4-39
RELEASE-NOTES-U6-en-
RELEASE-NOTES-U6-en- o bonobo-activation-2.2.2-1 => bonobo-activation-2.2.2-1.2E
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: autofs4: How to confg TCP 1st choice, not UPD?
2006-11-01 22:48 ` autofs4: How to confg TCP 1st choice, not UPD? Risto Bell
@ 2006-11-01 23:14 ` Jeff Moyer
0 siblings, 0 replies; 2+ messages in thread
From: Jeff Moyer @ 2006-11-01 23:14 UTC (permalink / raw)
To: Risto Bell; +Cc: autofs
==> Regarding Re: [autofs] autofs4: How to confg TCP 1st choice, not UPD?; Risto Bell <reb@cypress.com> adds:
reb> Using RedHat Enterprise 3 update 6. Release notes consistent with
reb> my experience that UDP is the default. Perhaps RedHat changed the
reb> default autofs4's natural TCP back to UDP? In any event, question
reb> remains: There isn't an simple config by which to control it (prefer
reb> TCP with timeo=600,retrans=2 1st), it has to be compiled into it?
The default is not embedded in autofs, it's part of the mount command
and/or the kernel mount implementation. My system is RHEL 5 beta, so much
newer than yours.
reb> Also could use multi-mounts? List two alternative mounts, same but
reb> one TCP and one UPD, so the 1st (TCP) will be attempted, and only if
reb> it fails (server doesn't support TCP) will it fall to 2nd (UDP)?
reb> This would still need programs/scripts but would save rcpinfo process,
reb> and achieve slight speed-up.
Please don't do this. It is not the correct way to go about doing things,
and it will not work. The reason is that when trying a replicated server,
if we fail to mount we mark the server as down. So, your subsequent
request will also fail.
-Jeff
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2006-11-01 23:14 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <mailman.122.1162331942.2205.autofs@linux.kernel.org>
2006-11-01 22:48 ` autofs4: How to confg TCP 1st choice, not UPD? Risto Bell
2006-11-01 23:14 ` Jeff Moyer
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.