GPFS webinar: day 2
Oct. 14th, 2011 02:57 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
sawylMore GPFS training, this time focused on troubleshooting and problem diagnosis. Most of the information was familiar — check the network connectivity, query the waiters, test the health of the disc subsystem — but it was useful to see the information packaged up as a series of case studies and to see the commands in use.
Whenever I get asked to look into a GPFS performance issue, I tend to look at each end of the problem — the application performance, then the disc and network performance — and skip GPFS in the middle, just because the default logs are both spectacularly unhelpful and distributed over every node in the cluster. So having a chance to see IBM's diagnostic methods in action, to see just how much information is available via the
Whenever I get asked to look into a GPFS performance issue, I tend to look at each end of the problem — the application performance, then the disc and network performance — and skip GPFS in the middle, just because the default logs are both spectacularly unhelpful and distributed over every node in the cluster. So having a chance to see IBM's diagnostic methods in action, to see just how much information is available via the
mmdiag command.
