I/OTest script to check if the disk I/O is causing slow performance
Slow VM Performacne, use IOTest to see if the disk IO is the culprit
This script will test the disk IO by copying 500Mb of data using the same block size as eDir uses and with the same api eDir uses “fdatasync”.
This writes 500 Mb of data each iteration to the iotest.log in the dib directory, usually the /var/opt/novell/eDirectory/data/dib/
It will overwrite the previous data in the iotest.log each time it runs. Anything under 100 MB/s is a concern and will cause slowness for eDirectory and possible memory build up. IO causes a bottleneck for events to be written to disk. A build up of memory by ndsd can cause a ndsd to take all available memory (both virtual and resident) causing ndsd to core.
If slow IO writes are seen with the iotest script begin the process of adding hard drives and reducing the… Continue reading
VMWare and best practices for Timekeeping
Lately I’ve been getting a lot of requests for timekeeping best practices for VMWare
VMWare has a great support article on this. Below are the SLES recommendations. The article can be found here
| SLES 11 (All updates) | No additional kernel parameters required. | |
| SLES 10 SP4 on ESX 5.0 and later | clock=pmtmr hpet=disable |
|
| SLES 10 SP4 on ESX 4.x | Use a VMI enabled kernel. | |
| SLES 10 SP3 on ESXi 5.0 | clock=pmtmr hpet=disable |
|
| SLES 10 SP3 on ESX 3.5 and 4.x | Use a VMI enabled kernel. | |
| SLES 10 SP3 on ESX 3.0.x and earlier | clock=pmtmr hpet=disable |
|
| SLES 10 SP2 on ESXi 5.0 | clock=pmtmr hpet=disable |
|
| SLES 10 SP2 on ESX 3.5 and 4.x | Use a VMI enabled kernel. | |
| SLES 10 SP2 on ESX 3.0.x and earlier | clock=pmtmr hpet=disable |
|
| SLES 10 SP1 | clock=pmtmr hpet=disable |
|
| SLES 10 | clock=pmtmr hpet=disable |
|
| SLES 9 (All updates) | clock=pmtmr hpet=disable |
|
| SLES 8 | No additional kernel parameters required.… Continue reading |