On Sat, 20 May 2006 09:04:40 PDT, Barton Robinson <[EMAIL PROTECTED] ftware.com> wrote:
>Alan, one command to collect raw data: >"SMSG ESASERVE RECORD DISK CONFIG" >gives ibm exactly what they want, >"SMSG ESASERVE RECORD NONE" stops recording raw data. >zero need for an added user or a lot of work. Do you think >we get test data from MONWRITE? I was going to leave this off a public forum -- but you started it. They couldn't get it to work. They dinked around with ESAMON for a week. We sent all the data they collected in to IBM, and NONE of it was the data IBM had asked for. Some of it was missing even basic data. IBM asked us to use MONWRITE instead of ESAMON to collec t the data. 1. They forgot to include the CONFIG option. Why the **** isn't CONFIG th e default? Raw data without CONIFG data is pretty much worthless. 2. The segment sizes weren't big enough. I tried running the tool you sup ply, but it appears to be out-of-date. So I made them 64M each and that problem went away. 3. The automation in ESASERVE kept overriding the MONITOR commands they i ssued, so they didn't get the I/O data that IBM asked for. They didn't know how to overr ide it. 4. On the last try, ESAWRITE crashed. I asked them to send you the dump b ut instead they decided to install the latest service. 5. They gave up on ESAWRITE and went back to MONWRITE. Either your product is very buggy (which is what they tell me) or it is t oo hard to set up and use for the people who are now in charge of it. I like to think I could fix this, but I'm not allowed to. Why don't I tak e care of this product? In a word, politics. I'm not saying PerfKit is better than ESAMON. (I doubt it.) It's just a l ot easier to collect raw monitor data with just MONITOR commands and MONWRITE.