I can see different time to respond when using large ping buffer size for one of our HP Blade servers. When using default buffer size, time to respond is the same on all servers.
As a first step to troubleshoot, I compared two servers in the same c7000 enclosure, where one server has the issue and other one does not.
Both servers are Windows 2k8 R2 BL490c Gen 7 servers, in the same C7000 enclosure, using same enclosure network module (HP 1/10Gb VC-Enet Module)
When pinging servers using large ping buffer for server1, I get consistent 1-2 ms response time. For server2, I get 3-4 ms :
ping -l 65500 server1
Pinging server1 [10.100.100.2] with 65500 bytes of data:
Reply from 10.100.100.2: bytes=65500 time=1ms TTL=127
Reply from 10.100.100.2: bytes=65500 time=2ms TTL=127
Reply from 10.100.100.2: bytes=65500 time=1ms TTL=127
Reply from 10.100.100.2: bytes=65500 time=1ms TTL=127
ping -l 65500 server2
Pinging server2 [10.100.100.3] with 65500 bytes of data:
Reply from 10.100.100.3: bytes=65500 time=3ms TTL=127
Reply from 10.100.100.3: bytes=65500 time=4ms TTL=127
Reply from 10.100.100.3: bytes=65500 time=4ms TTL=127
Reply from 10.100.100.3: bytes=65500 time=4ms TTL=127
This is consistent if I ping from different sources, i.e. from blades in oher enclosures
Both servers are using 2x bonded HP NC553i Dual Port FlexFabric 10Gb adapters
I can see that blade and Network adapter firmware on both blades is slightly different - it is actually newer on the server2
Other difference is that they are on the different VLAN, however we don't see any delay for blades in the same vlan on other enclosures.
I have checked that adapters are set to autonegotiate on both servers.
Where should I look to troubleshoot first? I would try to avoid firmware update as much as I can at this point.
0 Answers