|
in DDN-554 we need quite amount of data to analyse performance degradation. this isn't the only case we such a data is needed. every time we ask customer to collect this data, explain how to do so, etc.
it makes sense to have a script doing this.
ideally it should be a single command spawning all (or specified) servers.
yes another useful feature would be to show some data online, so that user could watch how servers are doing in different areas. for example, read/write rates for every OSS/OST - then user can easily identify underperforming target. similar mode can be used to monitor other stats (IO sizes, IO times, avg.seek, etc).
|