Error 2022 messages

[mark-lee]

Before I start, I have been in contact with Microsoft about this error and they have mentioned several patches or hot fixes. However our contention on the issue is based on a simple premise:

- EMC claims Mirrowview has NO IMPACT on host or host processing. Mirrorview is an array-to-array architecture. Since we do not encounter this error 2022 at any other time (with full 1000 user access + full backups occurring in the middle of a production day) other than when Mirrorview is active, it leads us to believe that the hosts are being impacted either by 1) mirrorview 2) the Cisco switching fabric is being overwhelmed or saturated by mirrorview requests 3) not enough bandwidth available for mirrorview to send and receive request to the remote system or possibly 4) disk contention. The result is the server when trying to access disk or network resources simple balks and returns the "error 2022 cannot establish connections after XXX tries". The incident occurs for a very short (1-3 minute) duration, clears, then is fine until the next occurrence.

Backtracking on when the problem started to occur several things happened at once (of course!):

1) The bandwidth between the local and remote 9216s was reduced from 500mb/sec to 256mb/sec.

2) MacAfee antivirus was updated

Between then and now, we have applied 1x MS HOT fix but they now suggest another. I really appreciate your assistance in this matter. We are hoping this is a tuning parameter somewhere in the configuration of the operating system, fabric, arrays, Emulex HBAs, etc.

Regards,

Mark

[tblancha]

I cannot speak for the error since it showing up on a MS server. But with regards to the MDS switch, depending on how often this occurs, it may be possible to do a fcanalyzer brief to show the interaction between the server and switch. Also, a span session with a port analyzer would be the best thing to show the frames in and out of the port either to the server. If you do a show log, is the port going offline? In the show interfaces, is there a signficant amount of NOS, LOS, or OLS? Are these numbers getting larger or do they stay the same from before the error message and then check after.