Strange network issues
alienn
37 Posts
February 12, 2020, 9:13 amQuote from alienn on February 12, 2020, 9:13 amHi,
I have a cluster consisting of three identical nodes. Each node feature two onboard 10g nic ports (Intel X722) and one dual port Broadcom 10g nic (BCM57840 NetXtreme II). One port of each type is connected to a Unifi 16XG 10g switch.
Now I have packet errors on the link of the onboard nics. I already replaced the cable and switch ports and switches but the packet errors follow the onboard nics.
I don't think that all of my three hosts have defective mainboards. 🙂
In the past I already experienced strange issues with nics made by Intel and Linux especially in the context of tcp segment offloading and such.
Does anybode else see this behaviour and or has an idea how to debug this kind of problem?
Cheers,
Nicki
Hi,
I have a cluster consisting of three identical nodes. Each node feature two onboard 10g nic ports (Intel X722) and one dual port Broadcom 10g nic (BCM57840 NetXtreme II). One port of each type is connected to a Unifi 16XG 10g switch.
Now I have packet errors on the link of the onboard nics. I already replaced the cable and switch ports and switches but the packet errors follow the onboard nics.
I don't think that all of my three hosts have defective mainboards. 🙂
In the past I already experienced strange issues with nics made by Intel and Linux especially in the context of tcp segment offloading and such.
Does anybode else see this behaviour and or has an idea how to debug this kind of problem?
Cheers,
Nicki
Strange network issues
alienn
37 Posts
Quote from alienn on February 12, 2020, 9:13 amHi,
I have a cluster consisting of three identical nodes. Each node feature two onboard 10g nic ports (Intel X722) and one dual port Broadcom 10g nic (BCM57840 NetXtreme II). One port of each type is connected to a Unifi 16XG 10g switch.
Now I have packet errors on the link of the onboard nics. I already replaced the cable and switch ports and switches but the packet errors follow the onboard nics.
I don't think that all of my three hosts have defective mainboards. 🙂
In the past I already experienced strange issues with nics made by Intel and Linux especially in the context of tcp segment offloading and such.
Does anybode else see this behaviour and or has an idea how to debug this kind of problem?
Cheers,
Nicki
Hi,
I have a cluster consisting of three identical nodes. Each node feature two onboard 10g nic ports (Intel X722) and one dual port Broadcom 10g nic (BCM57840 NetXtreme II). One port of each type is connected to a Unifi 16XG 10g switch.
Now I have packet errors on the link of the onboard nics. I already replaced the cable and switch ports and switches but the packet errors follow the onboard nics.
I don't think that all of my three hosts have defective mainboards. 🙂
In the past I already experienced strange issues with nics made by Intel and Linux especially in the context of tcp segment offloading and such.
Does anybode else see this behaviour and or has an idea how to debug this kind of problem?
Cheers,
Nicki