[lustre-discuss] lightweight RPC monitoring for portal RPC daemon?
ewahl at osc.edu
Mon Aug 24 12:24:47 PDT 2015
Anyone have a handy way to do some lightweight RPC monitoring for the portal RPC daemon? (ptlrpcd) I'm hoping someone has rigged something up for debugging before and can share. We're seeing some odd evicts/'stuck cpu until crash' issues that we'd like to take a closer look at. Lustre 2.5.x
Hoping to find something that stands out as to what causes the reconnects/evictions until we can recreate it. Expecting to find a user doing something degenerate but low bandwidth that hits a new LBUG.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the lustre-discuss