xulpher
1 Copper

Possibly an Atmos bug?

I've only been able to reproduce this on a specific file.  I don't really want to post it on here, but I can email it if I need to for those that want to do testing.  I'm uploading a compressed file via atmos_c (curl).  Every file works as expected except for the one in question.  When I retrieve the file back from the server is has the following text prepended to the data:

da6
100000
In order to rule out possible upstream bugs I ran a tcp dump of all http traffic and verified that the file was sent correctly, and then did a tcp dump when retrieving to verify that it wasn't just curl misinterpreting a header.  Another interesting thing to note is that this only appears to happen if the file is uploaded to completion.  If the file isn't completely stored and is retrieved then the initial part of the file I get is correct so perhaps this has to do with the size of the file?

Here is the offending incoming header/bad data:

Server: nginx

Date: Mon, 13 Sep 2010 20:33:56 GMT

Content-Type: application/octet-stream

Transfer-Encoding: chunked

Connection: keep-alive

x-emc-policy: Plus

x-emc-listable-meta: x-uid=1000, x-gid=100, x-permissions=33188, x-compression=1

x-emc-meta: atime=2010-09-13T20:09:18Z, mtime=2010-09-13T20:09:18Z, ctime=2010-0

9-13T20:09:18Z, itime=2010-09-13T20:05:42Z, type=regular, uid=<snip>, gid=apach

e, objectid=4b0ab729a1230da704b0abf13c86de04c8e841637bb5, objname=<snip>, size

=4701488, nlink=1, policyname=Plus

x-emc-useracl: 4015050=FULL_CONTROL

x-emc-groupacl: other=NONE

Transfer-Encoding: chunked

da6

100000

<start of actual file data>
The dump of the upload does not reveal da6/100000 anywhere.

0 Kudos
2 Replies
scot4
1 Copper

Re: Possibly an Atmos bug?

You could try putting the object in question up using AtmosFox http://www.suchisoft.com/atmosfox/AtmosFox.xpi or the simple obect Browser https://community.emc.com/docs/DOC-509 to (hopefully) narrow down the issue.

you can also send me the object in question at junkin_scot [ at ] emc.com and I can look into it

0 Kudos
hdawood
1 Copper

Re: Possibly an Atmos bug?

I am interested in the compressed file if you don't mind sharing it. I have a virtual environment setup so it's easy to see what happens on both ends of the spectrum (sending and receiving).

0 Kudos