On 10/16/2016 02:40 PM, Ankireddypalle Reddy wrote:
The encryption xlator is the last one before posix and it’s here that
the data is getting encrypted. When the data is read back the
encrypted data is returned. Decryption is supposed to happen in read
callback which does not seem to be happening. The fact that encrypted
data is getting returned indicates that data in turn is getting
returned from the posix/underlying fs layer. Is it possible that data
be returned by reading from the underlying fs by any translator other
than posix.
No, rot13_readv() just winds it to the xlator below (posix) and decrypts
the read in the cbk. Have you tried disabling the perf xlators on the
client like I suggested? You can see if client3_3_readv() is hit on the
client (mount) process when you read the file. If it is and despite it,
posix_readv is not hit on the brick, then something fishy is going on.
-Ravi
Thanks and Regards,
Ram
*From:*Ravishankar N [mailto:[email protected]]
*Sent:* Sunday, October 16, 2016 12:19 AM
*To:* Ankireddypalle Reddy; [email protected]
*Subject:* Re: [Gluster-users] rot-13 Translator query
On 10/15/2016 08:22 PM, Ankireddypalle Reddy wrote:
Hi,
I am trying to follow the below document for developing a
translator.
https://github.com/gluster/glusterfs/blob/master/doc/developer-guide/translator-development.md
I’ve created a replica volume and modified the vol file to
include rot-13 translator. Below is the snippet from vol file.
volume myvol-posix
type storage/posix
option volume-id b492191e-77a5-4fc3-9394-49218e36dae2
option directory /brick1/repli
end-volume
volume *myvol-rot13*
type encryption/rot-13
subvolumes *myvol-posix*
end-volume
volume myvol-trash
type features/trash
option trash-internal-op off
option brick-path /brick1/repli
option trash-dir .trashcan
subvolumes *myvol-rot13*
end-volume
…
The writes are getting intercepted by the translator and the file
is getting encrypted. But the reads don’t seem to be getting
intercepted by the translator. I tried setting break point in the
posix_readv function and attach the brick daemons to gdb. But
posix_readv does not seem to be getting called on the brick daemon
and the read completes on the application side.
Can someone please explain how the reads are getting serviced here
without hitting the posix layer.
It could be due to client side caching. I usually disable all
performance xlators (write-behind, read-head, io-cache, stat-prefetch,
quick-read, open-behind) when I want to remove caching effects while
debugging. drop-caches also helps.
HTH,
Ravi
Thanks and Regards,
Ram
***************************Legal Disclaimer***************************
"This communication may contain confidential and privileged material
for the
sole use of the intended recipient. Any unauthorized review, use or
distribution
by others is strictly prohibited. If you have received the message by
mistake,
please advise the sender by reply email and delete the message. Thank
you."
**********************************************************************
_______________________________________________
Gluster-users mailing list
[email protected] <mailto:[email protected]>
http://www.gluster.org/mailman/listinfo/gluster-users
***************************Legal Disclaimer***************************
"This communication may contain confidential and privileged material
for the
sole use of the intended recipient. Any unauthorized review, use or
distribution
by others is strictly prohibited. If you have received the message by
mistake,
please advise the sender by reply email and delete the message. Thank
you."
**********************************************************************
_______________________________________________
Gluster-users mailing list
[email protected]
http://www.gluster.org/mailman/listinfo/gluster-users
