osd/ReplicatedBackend: do not set omap header if it is empty

[fangyuxiangGL]

In replicatedpg backfill, objects have no omap header or keys previously, but an empty omap header will be inserted when they are pushed to replica, which leads leveldb performance degrade seriously when cluster has huge number of objects.
so my opinion is that we does't insert omap header if is size is zero when handle PushOp in recovery.
Signed-off-by: fangyuxiang fang.yuxiang@eisoo.com

[fangyuxiangGL]

Signed-off-by: fangyuxiang fang.yuxiang@eisoo.com

[hjwsm1989]

you should add the "Signed-off-by: xxx" line to your commit meesage,
check other PRs

[hjwsm1989]

and dont close/reopen the pr,
you can use "git push -f" to force update

[fangyuxiangGL]

@hjwsm1989 thx

[fangyuxiangGL]

@hjwsm1989 could you tell me how can i invite a reviewer for this pr? thx

[fangyuxiangGL]

@liewegas

Hi, sage

could you review for me? first PR, thx :-)

[tchaikov]

could you prefix the title of your commit message with the subcomponent your are changing ? see https://github.com/ceph/ceph/blob/master/SubmittingPatches.rst#3-describe-your-changes, in this case, it would be "ReplicatedBackend:".

and you could shorten the title of your commit message to something like:

ReplicatedBackend: do not set omap header if it is empty

and put more details in the body of the commit message.

also, you might want to add brackets around your mail address in your commit message like

Signed-off-by: fang yuxiang fang.yuxiang@eisoo.com

"git commit -s" will do this for you.

[fangyuxiangGL]

@tchaikov ok, thanks for your remind

[fangyuxiangGL]

@tchaikov amended the commit message as the rule, thx