I was gsyncing 200gb file from master to slave - initially it sync'ed 3gb to slave and for almost 1 hr slave shows file size as 3gb and then it showed the size as 45gb - Now for more  than 2 hrs file size stays at 45gb. but master and slave process are still syncing.

It would be better to add status of which file is currently getting synced to slave and (if possible) how much of it already gsynced to slave.
seems like rsync itself doesn't have such option. something like lsof + rsync could be used to get such info.

(In reply to comment #1)
> I was gsyncing 200gb file from master to slave - initially it sync'ed 3gb to
> slave and for almost 1 hr slave shows file size as 3gb and then it showed the
> size as 45gb - Now for more  than 2 hrs file size stays at 45gb. but master and
> slave process are still syncing.
> 
> It would be better to add status of which file is currently getting synced to
> slave and (if possible) how much of it already gsynced to slave.
> seems like rsync itself doesn't have such option. something like lsof + rsync
> could be used to get such info.

Vijay have already proposed this idea. I suggested to postpone it to a later time; being able to retrieve the very basic info of what synchronizations are set up and are functional is part of the soundness of the service. So we try to get this part right first, then we can come to "bit sausages" (do you use this slang for progress bar ;) ?)

PATCH: http://patches.gluster.com/patch/6866 in master (mgmt/glusterd: Implementation of volume gsync status [master [slave]])

PATCH: http://patches.gluster.com/patch/6874 in master (DHT: Add xlator-option assert_no_child_down)

Kaushik,

please don't mark it resolved until the required further changes get in.

(In reply to comment #5)
What are the further changes for this? Any of that getting in to 3.2.0? If not, we can move the target milestone.

some re-factoring in code is required, will send a patch out today.

PATCH: http://patches.gluster.com/patch/6929 in master (mgmt/glusterd: unify the geo-replication status rpc messages.)

PATCH: http://patches.gluster.com/patch/6933 in master (cli: UI cleanup for geo-replication command)

I reopen it to make a placeholder for an enhancement which I feel desirable regarding the status.

Along with sessions, either the sugared slave url the user passed on the command line, or a convenient semi-desugared url (like the one you see in gsyncd logs when it clams "syncing from a to b") should be recorded and be used in output.

Having been exposed to it personally, I find it rather unpleasant to see raw canonicalized urls with domain names forcibly resolved to ip addresses and the overwhelming presence of scheme specificators.

Needed changes in python and/or glusterd code are to be discussed, now just make a note so that not to forget of it.

*** Bug 3233 has been marked as a duplicate of this bug. ***

CHANGE: http://review.gluster.com/79 (Ie. instead of writing out the fully expanded canonical URL like) merged in master by Vijay Bellur (vijay)