Hi  Markos,

nope, it was not something related to the worker being restarted.

Anyway, I think I could find a solution... we are using monit to monitor
the status of the services, ftp service included.

FTP monitoring was always behaving weirdly (not matching the real status of
the service, for instance) so I thought about a possible "conflict" between
the two and I tried to switch it off. After two weeks everything is working
and I had not received any further complain from the customer so I guess
it's working now :)

To sum it up: looks like monit can break your FTP connection.

I'll update this post as soon as I have more positive feedback on this.

Thanks for your answer!
S.


On Sun, Mar 8, 2015 at 3:16 PM, Markos Gogoulos <[email protected]> wrote:

> Hi Simone,
>
> is there a chance that the worker restarts during the upload and this
> closes the connection? No idea if this could be related. Btw I just tried
> uploading 10 small videos (<10Mb) through ftp and they all got uploaded.
> Problem is they weren't picked by the transcodedaemon, something that Olof
> from Sweden has also reported recently. I had to restart the worker and
> press retranscode one or two times for them to be handled correctly by the
> transcodedaemon...
>
> Cheers
>
>
>
> On Mon, Feb 23, 2015 at 12:52 PM, Simone Orsi <[email protected]> wrote:
>
>> Hi,
>>
>> sometimes the FTP server that rely on worker instance is really unstable.
>>
>> I just tried to upload  video and while uploading it got 4 disconnections
>> that were fortunately handled well and the upload resumed from where it
>> stopped.
>>
>> After some minutes the connection went away again and didn't get up in
>> time, so that Filezilla's log said:
>>
>> Status: Delaying connection for 5 seconds due to previously failed
>> connection attempt...
>> Status: Resolving address of #######
>> Status: Connecting to #######:21...
>> Status: Connection attempt failed with "ECONNREFUSED - Connection
>> refused by server".
>> Error: Could not connect to server
>>
>> and ploneftp log said:
>>
>> []2.236.246.250:52364 Connected.
>> 2.236.246.250:52364 ==> 220 pyftpdlib 0.7.0 ready.
>> 2.236.246.250:52364 <== USER #####
>> 2.236.246.250:52364 ==> 331 Username ok, send password.
>> 2.236.246.250:52364 <== PASS ******
>> authenticating to Zope FTP server 0.0.0.0:8021
>>  with username #####
>> authenticated successfully with Plone FTP
>> 2.236.246.250:52364 ==> 230
>> 2.236.246.250:52364 <== OPTS MLST
>> type;perm;size;modify;unix.mode;unix.uid;unix.gid;
>> 2.236.246.250:52364 ==> 200 MLST OPTS
>> type;perm;size;modify;unix.mode;unix.uid;unix.gid;
>> 2.236.246.250:52364 <== CWD /
>> 2.236.246.250:52364 ==> 250 "/" is the current directory.
>> 2.236.246.250:52364 <== PWD
>> 2.236.246.250:52364 ==> 257 "/" is the current directory.
>> 2.236.246.250:52364 <== TYPE I
>> 2.236.246.250:52364 ==> 200 Type set to: Binary.
>> 2.236.246.250:52364 <== PASV
>> 2.236.246.250:52364 ==> 227 Entering passive mode
>> (192,107,92,224,201,23).
>> 2.236.246.250:52364 <== STOR millumino-test-upload.mp4
>> 2.236.246.250:52364 ==> 150 File status okay. About to open data
>> connection.
>> []@127.0.0.1:49417 Disconnected.
>> ZopeHandler got a closed connection
>>
>> So, it seems that the zope instance (the worker in this case, as per
>> plumi default setup) is closing the connection on the FTP port.
>>
>> We tried to increment the number of threads of this instance from 2 to 10
>> but it looks like it has no effects.
>>
>> What could cause this weird behavior in your experience?
>> What can we do to debug this or to tweak our server setup in order to
>> avoid this?
>>
>> Thanks in advance for any pointers :)
>>
>> Cheers,
>> S.
>>
>>
>
>
> --
> https://unweb.me
> state of the art information systems
>
_______________________________________________
Discuss mailing list
[email protected]
http://lists.plumi.org/listinfo/discuss

Reply via email to