Load-Balancing on an Etherchannel (LAG - Port-Channel)

[CliveG]

Can anyone explain to me, if the default hash algorithm on an etherchannel is "src-dst-ip" and this should load-balnce across the two ports equally (as it utilises the Source XOR Destination IP address for know IPs), why I see the following traffic usage:

Port-channel 22 - Interfaces te2/5/7 and te2/5/7

te1/5/7 - txload 95/255  and te2/5/7 - txload 1/255

The above results could be that te1/5/7 is running at almost 100% (240/255) and stil te2/5/7 would show a txload of about 4 or 5 (5/255).

The configuration is a basic default config of the port-channel with the IP Address and the physical interfaces are just set with basic confiig and the channel-group 22 active commands.

The transceiver information for the 2 interfaces are as follows:

Optical Optical
Temperature Voltage Current Tx Power Rx Power
Port (Celsius) (Volts) (mA) (dBm) (dBm)
---------- ----------- ------- -------- -------- --------
Te1/5/7 29.5 3.27 4.6 -7.4 -6.4

Optical Optical
Temperature Voltage Current Tx Power Rx Power
Port (Celsius) (Volts) (mA) (dBm) (dBm)
---------- ----------- ------- -------- -------- --------
Te2/5/7 46.8 3.27 23.9 -7.1 -8.8

The only thing that is different is the temparature of the transceiver, but I do not believe this is a hardware fault causing this. It looks more like the load-balancing on the etherchannel.

Any recommendations please?

I have read through all the docs. Maybe the adaptive has algorithm could help but I cannot actually shut this channel down as it is used 100% of the time, day and night.

[MHM Cisco World · Accepted solution]

etherchannel is "src-dst-ip" and this should load-balance across the two ports equally ??
that not right, 
if the source and destination is same then always the traffic will select one port-member of PO 
instead use scr-dst-ip + L4 port 

[CliveG]

Thank you. This is what I was told but what you have said is what I am seeing and is correct.

So, my question is, as this link is always utilised for streaming (therefore IGMP and the IPs will be the same) 24 hours a day and we have no redundancy in place (this is being rectified, but not quite yet), to stop the first port being utilised so heavily (because of the default hash algorithm) is there a way I can change this to adaptive without having to shut the channel down and back up again?

And also, will adaptive give me the equal load-balancing I would like or would this affect the streams?