We have HP Gen9 with Emulex 10Gb (Driver v10.5.121.7) running latest ESXi 6.0 and experience high latency of above 50ms to 1.5 seconds with LDAP queries between Windows DC to DC or VM to DC. However, moving the DC to a ESXi 5.5/6.0 with 1Gb network latency is down to below 50ms. If you just ping the two servers, ping response is <1ms.
Anyone else experiencing such issue? It appears to happen with ESXi 6.0 with 10Gb Networki. ESXi 5.5 on 10Gb appears to work fine.
>50ms up to 1.8 seconds in query time
ESXi6 DC VM to ESXi DC VM on the same host, 10Gb DSwitch
ESXi6 DC VM to ESXi DC VM on different host, 10Gb DSwitch
<50ms
Physical DC to Physical DC
ESXi5.5/6.0 DC VM to ESXi5.5/6.0 DC VM on 1Gb Network (Ping is <1ms, FIle copy is > 100MB/sec)
I've checked packet drops with esxtop with network view, no packet drops.
I used the following script for measurement which does 10 queries for a user and takes the average time.
The result is fairy consistent.
when bare metal servers and vm windows 2012 are involved then try disalble rsc in the vm
We have a similar issue..
HP Proliant BL460 Gen9 with 650FLB Driver: 10.5.121.7 Firmware: 10.5.160.
Regarding Firmware 10.5.160.1, there is an HPE advisory which describes why to use this firmware:
https://h20564.www2.hpe.com/hpsc/doc/public/display?docId=emr_na-c04855352
Latest Vmware 6.0 and HP Patches ...
Issue:
VM´s MS AD Server and Excange (MSHUB Servers) - looking at the performance monitor in windows (Exchange Ldap queries)
when AD and HUB are on the same esxi - no problem - latency >5ms
when AD and HUB are on different esxi (no matter in same or different enclosure) - problem - lateny <500ms = exchange is slow
testing also with "psping" and found out that other windows vm´s are also affected - doesn´t see this issue onany linux vm
testing different hardware versions (9,10 and 11); testing with different vm-tools version, testing also with differnet network adapert (e1000) = no change do the issue
also no problem on the network (rx, drops etc.)
case at HP and Vmware is open ...
which servers do you use?
I have a series of Dell and SGI servers. All of them are using Broadcom NetExtreme II BCM57810 10gbps NICs.
Tests between various hosts/VMs and bare metal boxes net below 25ms in most cases average is about 5-10ms.
It doesn't seem to be an ESXi 6 issue with 10gbps cards.
Could be an HP driver issue.
when bare metal servers and vm windows 2012 are involved then try disalble rsc in the vm
Windows 2012 R2 Domain Controllers and Exchange 2010 Windows 2008 R2 servers.
Great for VMWare support, ticket has been opened for 1 month now and nothing.....cricket.......
Nice, I've disabled and tested all other settings except the PSC!
We have disabled PSC across the board on the ESXi Host according to VMWare document. LDAP is great.
Also, this impacted our View environment. However disabling LRO on host causes PSOD with Teridici...daily....AWESOME!
Hi All,
we have the similar issue on our environment as well and during the course of troubleshooting i have disable the RSS and RSC on IPv4 and IPv6 and the slow LDAP query response issue was sorted for us.
I hope there will be a VMware KB article on this soon as almost every application requires an ldap response and this has slowed down the overall performance of the VMs.
Hi All,
we have the similar issue on our environment as well and during the course of troubleshooting i have disable the RSS and RSC on IPv4 and IPv6 and the slow LDAP query response issue was sorted for us.
DisableNetadapterRSC -Name "Network Adapter name" -IPv4
DisableNetadapterRSC -Name "Network Adapter name" -IPv6
I hope there will be a VMware KB article on this soon as almost every application requires an ldap response and this has slowed down the overall performance of the VMs.