PowerVM

Hello,

I wonder if anyone from group experienced vNIC failover which takes a lot of time? We were testing vNIC for last few years with IBM i clients but apparently the technology does not offer an enterprise standard. We have seen several times that the failover takes too long (from 30 seconds to few minutes) which causing many of TCP jobs to crash. 

We are not able to simulate the scenario to provide IBM enough logs, but we have seen long failover if there is an outage on a LAN switch or when SRIOV firmware upgrade is happening. If we initiate a failover manually, it is smooth and fast. 

Unfortunately, this unpredicted behavior is stopping us with promotion vNIC as main network virtualization engine. 

Does anyone else experienced same behavior on the other operating systems?  I've seen this document which seems to document that there is an issue somewhere. 

Why vNIC failover/failback takes long time and uses high CPU on VIOS during failover/failback?

Bart G,

Hello,

I wonder if anyone from group experienced vNIC failover which takes a lot of time? We were testing vNIC for last few years with IBM i clients but apparently the technology does not offer an enterprise standard. We have seen several times that the failover takes too long (from 30 seconds to few minutes) which causing many of TCP jobs to crash. 

We are not able to simulate the scenario to provide IBM enough logs, but we have seen long failover if there is an outage on a LAN switch or when SRIOV firmware upgrade is happening. If we initiate a failover manually, it is smooth and fast. 

Unfortunately, this unpredicted behavior is stopping us with promotion vNIC as main network virtualization engine. 

Does anyone else experienced same behavior on the other operating systems?  I've seen this document which seems to document that there is an issue somewhere. 

Why vNIC failover/failback takes long time and uses high CPU on VIOS during failover/failback?

Bart G,

------------------------------
Bartlomiej Grabowski
------------------------------

Hello Bart,

I experience the same thing for a client running IBM i.  The Vnic failover takes up to 1 minute.  The client opened a support ticket and as of today, there is no way to solve this failover latency.  

Hello,

I wonder if anyone from group experienced vNIC failover which takes a lot of time? We were testing vNIC for last few years with IBM i clients but apparently the technology does not offer an enterprise standard. We have seen several times that the failover takes too long (from 30 seconds to few minutes) which causing many of TCP jobs to crash. 

We are not able to simulate the scenario to provide IBM enough logs, but we have seen long failover if there is an outage on a LAN switch or when SRIOV firmware upgrade is happening. If we initiate a failover manually, it is smooth and fast. 

Unfortunately, this unpredicted behavior is stopping us with promotion vNIC as main network virtualization engine. 

Does anyone else experienced same behavior on the other operating systems?  I've seen this document which seems to document that there is an issue somewhere. 

Why vNIC failover/failback takes long time and uses high CPU on VIOS during failover/failback?

Bart G,

------------------------------
Bartlomiej Grabowski
------------------------------

Hello,

I wonder if anyone from group experienced vNIC failover which takes a lot of time? We were testing vNIC for last few years with IBM i clients but apparently the technology does not offer an enterprise standard. We have seen several times that the failover takes too long (from 30 seconds to few minutes) which causing many of TCP jobs to crash. 

We are not able to simulate the scenario to provide IBM enough logs, but we have seen long failover if there is an outage on a LAN switch or when SRIOV firmware upgrade is happening. If we initiate a failover manually, it is smooth and fast. 

Unfortunately, this unpredicted behavior is stopping us with promotion vNIC as main network virtualization engine. 

Does anyone else experienced same behavior on the other operating systems?  I've seen this document which seems to document that there is an issue somewhere. 

Why vNIC failover/failback takes long time and uses high CPU on VIOS during failover/failback?

Bart G,

------------------------------
Bartlomiej Grabowski
------------------------------

It looks the problem is well know. I wonder why the tech support does not reply - "known behavior without a permanent fix". Instead they are playing the game that we are the first company to complain, and asking for collecting never ending logs. 

I should be in IBM Rochester in few weeks, do you mind to share PMR numbers? Maybe we can get a bigger attention from the management. 

------------------------------
Bartlomiej Grabowski
------------------------------

Original Message:
Sent: Mon January 15, 2024 10:38 AM
From: Bartlomiej Grabowski
Subject: vNIC - failover takes long time

Hello,

I wonder if anyone from group experienced vNIC failover which takes a lot of time? We were testing vNIC for last few years with IBM i clients but apparently the technology does not offer an enterprise standard. We have seen several times that the failover takes too long (from 30 seconds to few minutes) which causing many of TCP jobs to crash. 

We are not able to simulate the scenario to provide IBM enough logs, but we have seen long failover if there is an outage on a LAN switch or when SRIOV firmware upgrade is happening. If we initiate a failover manually, it is smooth and fast. 

Unfortunately, this unpredicted behavior is stopping us with promotion vNIC as main network virtualization engine. 

Does anyone else experienced same behavior on the other operating systems?  I've seen this document which seems to document that there is an issue somewhere. 

Why vNIC failover/failback takes long time and uses high CPU on VIOS during failover/failback?

Bart G,

------------------------------
Bartlomiej Grabowski
------------------------------

It looks the problem is well know. I wonder why the tech support does not reply - "known behavior without a permanent fix". Instead they are playing the game that we are the first company to complain, and asking for collecting never ending logs. 

I should be in IBM Rochester in few weeks, do you mind to share PMR numbers? Maybe we can get a bigger attention from the management. 

------------------------------
Bartlomiej Grabowski

Original Message:
Sent: Mon January 15, 2024 10:38 AM
From: Bartlomiej Grabowski
Subject: vNIC - failover takes long time

Hello,

I wonder if anyone from group experienced vNIC failover which takes a lot of time? We were testing vNIC for last few years with IBM i clients but apparently the technology does not offer an enterprise standard. We have seen several times that the failover takes too long (from 30 seconds to few minutes) which causing many of TCP jobs to crash. 

We are not able to simulate the scenario to provide IBM enough logs, but we have seen long failover if there is an outage on a LAN switch or when SRIOV firmware upgrade is happening. If we initiate a failover manually, it is smooth and fast. 

Unfortunately, this unpredicted behavior is stopping us with promotion vNIC as main network virtualization engine. 

Does anyone else experienced same behavior on the other operating systems?  I've seen this document which seems to document that there is an issue somewhere. 

Why vNIC failover/failback takes long time and uses high CPU on VIOS during failover/failback?

Bart G,

------------------------------
Bartlomiej Grabowski
------------------------------