Thanks for the feedback!
Matt, I have already several others OPs traced, I was looking if this will
work or not, as you said I will try to trace more on beast and and try to
go deeper for other functions as well.
Once again thank you Matt Benjamin.
On Thu, 14 May 2020, 06:10 Matt Benjamin, <mbenjami(a)redhat.com> wrote:
I'm having trouble reading the labels clearly, I have to admit. Still,
this looks like a great start on "core" op processing.
I'm unsure of your available time. It would be super to work horizontally
across more ops. It also makes sense to go deeper--but especially, to be
able to jump across (I think) two domains that are off the left and right
edges. To the left, it seems you can go some ways into the HTTP front-end
(in that direction, I think we care about Beast and maybe not Civetweb so
much). Meanwhile, to the right, most ops are pushing down into Ceph RADOS
ops, and it would be particularly nice to trace across RGW high-level ops,
into the underlying RADOS ops that work against the Ceph cluster.
In fine, nice work.
On Wed, May 13, 2020 at 8:00 PM Abhinav Singh <singhabhinav0796(a)gmail.com>
I m implementing a jaeger tracing system in RGW, some images below
[image: RGW_DELETE_OBJ.png]For deleting a object
for getting the list of buckets in clusters
I have two tags in jaegerUI to filter various spans one is gateway(swift
or s3) and another is RGwOpeartion type like for example putting an object
has a name "RGW_OP_PUT_OBJ".
is this much detail sufficient? or should I go more deep into each
functions and try to trace those.
Also what should any improvement and changes I can make into this to make
this more developer friendly.
Dev mailing list -- dev(a)ceph.io
To unsubscribe send an email to dev-leave(a)ceph.io
Red Hat, Inc.
315 West Huron Street, Suite 140A
Ann Arbor, Michigan 48103