Hello,
The KB5052 from oct 2008 (now also listed as KB ID: 3011562 Version: 6.0) is more or less the only document on the NOW site you can find about the maxdirsize.
According the article, changing the maxdirsize has an performance impact, but its hard to quantify. This is rather vague description and doesn't tell much about the real impact you can suffer on the system.
I was wondering if somebody has some real life experience in their NetApp environment after changing the maxdirsize (or in combination with maxfiles) on a volume ?
We raised it on one of our volumes from the default 73400 to 199999 (so that would be 2,72% and less than 3% mentioned in the document). We have a FAS3240 metrocluster / FlashCache 512GB / ONTAP 8.0.2.P4. We see on various moments that processor 3 is at 100%, doing 70% kahuna and 30% wafl_exempt. During this periods we are having big latency problems on all protocols (FCP - iSCSI - NFS). The specific volume is accessed over NFS and has around 16million files in the volume. Since it growed over the last year due to migrations, we are not sure that this volume is the causer. We suspect it however, because we had a serious issue with it after a snapmirror update. The initialize went ok, but on first snapmirror update, the cpu run wild and serious latency was the outcome. After breaking the snapmirror again, everything became stable again. NetApp support investigated the issues, but couldn't conclude anything.
Regards,
Geert
