Is the aix mem_free report correct in ganglia 2.5.0? I have it showing results from 0gig utilized, then slowly incrementing over an hour or two up to 4gig utilized -- the max memory thats on the system.
I took the relevant code in gmond's machine.c and made a test file which calls knlist to get the offset into /dev/kmem, seeks there and reads 192 bytes into a struct vmker.. but then the values I read out of buf->freemem and buf->totalmem don't make sense. In fact the value in buf->freemem is increasing on a consistent basis. Even re-computing it with the getpageinfo() / 1024 multiplication used in the gmond source, doesn't help. This is what knlist returns by the way, for the kernelimages structure $ ./vminfo knlist returned 0 3715656 0 0 0 0 3718912 0 0 0 0 947504 0 0 0 0 0 0 0 0 0 And that first offset 3715656 is the seek value into /dev/kmem. Does that seem reasonable? I was going to investigate this further to see exactly what kind of values the gmond process is coming up with in the relavent sections of code, but I thought I'd ask here. Also, does anyone know if ibm has a library for 4.3 for the vmgetinfo function? Its also mentioned in the gmond 2.5.0 code, but as being only available for aix 5. If someone can show me the exact monitor or vmstat or svmon command that should exactly equate to the mem_free output that is supposed to come up in the graphs, please let me know. From what I can tell the free memory and avm memory on the machine are 2025 mb, and 1495 mb respectivly, with the vmstat fre and avm values being 518421 382895 to match that, as examples, and the gmond graph output is just showing mem_free usage climb up and up as described above and then cleared out from 4gig, down to 0gig. My sense now is that it looks like its just wrapping around. Lester
