remotefs-users

Re: [remotefs-users] RemoteFS send/recv timeout

[Aleksey T. <ale...@gm...>]

Hi,

The timeout is only used to detect connection loss and for connecting to
remote host. 15 seconds is just a reasonable value put there for defaults,
if you increase this value then no issues should occur except detecting
connection loss will take longer time.

After looking into debug log, i'm not entirely sure why setting longer
timeout helps to overcome this difficulty. It appears from log that server
is handling writes fine until it receive new incoming connection which
tries to write to the file it didn't open before. Previous connection was
writing to the file descriptor with the same number, but new connection
opens another file it seems. RFS server consider this as internal error
which shouldn't normally happen and aborts the connection.

I have to ask if you're using any special software for copying or auto
remounting, or something like that. This might help to debug this issue.

There is no any specific recommended options to mount RFS from docker
container, but i see you're using nodirect_io option which should be
impossible option for RFS client and it should be filtered out if you're
mounting it as in `mount /shared_folder`. I assume intention was to disable
"direct_io" option of FUSE, to my knowlege it's not enabled by default, so
no harm should be done, just wanted to point that out.

If you increase the timeouts then check that MD5 sum of transferred file is
correct, then this might be some network delays between container and host
OS, but i would need some additional information to reproduce this:

1. What host OS you're using (Linux dstribution)
2. How large file transfer should be to reproduce this
3. You're copying single file or multiple files
4. You're using `cp` or other copying software
5. Information on any other software which might be involved into
copying/networking

Thanks.


On Thu, Oct 8, 2015 at 5:59 AM, Kuan Po Huang <kp...@gm...> wrote:

> Hi all,
>
> After modifying the send/recv timeout from 15 to 180 secs, the
> disconnection issue does not occur so far.
>
> Why is timeout 15 secs?
> Any possible issue occurs when timeout is raised?
>
> Thanks,
> Kuan Po
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> remotefs-users mailing list
> rem...@li...
> https://lists.sourceforge.net/lists/listinfo/remotefs-users
>
>

[remotefs-users] RemoteFS send/recv timeout

[Kuan Po H. <kp...@gm...>]

Hi all,

After modifying the send/recv timeout from 15 to 180 secs, the
disconnection issue does not occur so far.

Why is timeout 15 secs?
Any possible issue occurs when timeout is raised?

Thanks,
Kuan Po

[remotefs-users] remotefs is disconnected unexpectedly

[Kuan Po H. <kp...@gm...>]

Hi all,

Attachment is rfsd debug log before remotefs client(or server) is
disconnected unexpectedly .

I use remotefs for sharing between host OS and docker container
Following are rfsd and rfs configurations:

    host OS side:
rfsd -e "/etc/rfs-exports" -a 172.17.42.1 -u root -s "/etc/passwd" -q

docker container side using /etc/fstab:
172.17.42.1:/shared_folder /shared_folder rfs
users,username=root,password=/shared/passwd,nodirect_io 0 0

When doing massive data transmission, forked child process is disconnected
without any error(no out of memory and disconnection log in
/var/log/syslog), also possibly i didn't monitor the right network
information

How can i spot the network issue?

Any recommended remotefs mount options for between host OS and docker
container?

Appreciate it,
Kuan-Po Huang

[remotefs-users] remotefs 0.12 released

[Aleksey T. <ale...@gm...>]

Changelog: http://remotefs.svn.sourceforge.net/viewvc/remotefs/tags/0.12/CHANGELOG?revision=986