Some IBM folks have made an initial stab at abstracting the NFS subsystem into proper systemtap probes, but it needs a review. It seems as if the current batch are an excessively shallow (mechanical, providing little useful abstraction) and may need to be drastically cut down. See the systemtap src/tapset/nfs* files.
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
Is this a new feature? Looks like it from the description, but not marked. How the testing will be performed, could you provide some testing tips?
This is not a new feature so much as a review and fixing of existing code, which steved kindly volunteered to do some time ago. Ideally, it would be tested by specifically written systemtap scripts that exercise the new tapset code, to trace or analyze NFS goings-on.
OK, sounds good. As soon as the test scripts is ready, qa_ack will be provided. Is IBM also willing to help with testing?
Steve, your people.redhat.com scripts could be embedded in systemtap. We can ship them as examples (which are at least compile-tested during the systemtap testsuite), and some of them could be put into the full execute-test suite too. So there's already a place to put them that will allow them to be kept up-to-date. If you like, I can do (or have someone on the team do) an initial import of your stuff into our repository, and thence you could maintain it directly.
That would be cool to have my scripts embedded, but maybe it would be a better idea if have some show me how to embed them so in the future I could add and subtract things in less of a disruptive way (at lease in theory ;-) )
Details coming in email.
We need exact details of what is been asked to do in this bug; in order, to assess the amount of work and who best to assign the work to. Are new scripts to be written and tested, or there are already scripts that need to be tested? Without this detail we don't really know what to do. The request for Info has been posted in this bug for a while, now.
I believe the idea was for me to go through the current nfs, nfsd, lockd, and rpc tap scripts to see how I could improve or update and make them more use able people in my line of work (i.e. kernel people). Unfortunately instead of development those, I ended up developing my own set of stap scripts which are at: git://fedorapeople.org/~steved/systemtap.git So what probably needs to happen is a selective merge of some of my script into the systemtap rpm. So I guess we could push this to 5.5 or even make it an Fedora bug in hopes of getting the merge into RHEL6
------- Comment From 2011-02-21 00:31 EDT------- Closing this bug as there is no activity going on past 1 year 8 months. Please feel free to re-open if you have any update. Thanks Muni
The nfs tapset has been updated and used over the last few releases. Any specific problems will get new bugzilla entries.