Weird error during export of node to another server

[Colin]

2016-10-04, 09:55:47 ANR8213E Socket 46 aborted due to send error; error 104. (SESSION: 40251, PROCESS: 197)
2016-10-04, 09:55:47 ANR0515I Process 197 closed volume 000173. (SESSION: 40251, PROCESS: 197)
2016-10-04, 09:55:47 ANR0724E EXPORT NODE: Processing terminated abnormally - transaction failure. (SESSION: 40251, PROCESS: 197)

Then the export completes with a state of failure.

I have seen this occur on several exports. At first I thought it was because I didn't have a second management class setup that the node was expecting to be on the new server. After fixing that still no go. Any ideas?

 

[marclant]

ANR8213E Socket 46 aborted due to send error; error 104.

That's a communication error between the two. http://www.ibm.com/support/knowledgecenter/SSGSG7_7.1.3/srv.msgs/ANR8213E_Linux.html
The error 104 is an OS error code, on Linux is means "Connection reset by peer" source: http://www.ioplex.com/~miallen/errcmp.html
Guessing you are using Linux as other OS don't have an error code 104.

Check the target server activity log for communication errors around 2016-10-04, 09:55:47.
Also check the OS logs on both source and target.
I'd suggest to check with your network admins, but they always say the network is good, so I'll leave that to your discretion.

 

[Colin]

Well I am lucky in that regard as I am also the network admin. And the network is good :3. Next time it happens I will check what the corresponding log is on the importing side and post it here. Stay tuned.

 

[Colin]

Hrmm. Same problem happened that kept me from getting the importing side's actlog. The importing TSM server seems to start having trouble with its actlog getting filled to the point that I can't even query it anymore. That's when the exports seem (to that server) seem to start failing and giving these transaction failures.

Some of the errors included:

ANR0157W
ANR0158W
ANR9999D_3095886799


I did some googling and this is the most detail I can find on it: http://www-01.ibm.com/support/docview.wss?uid=swg21657999


Unfortunately as it fills up the actlog so I can't even read from it I can't get details from the server itself.

I am in the process of trying to export/import a huge node (about 50TB) when this seems to happen. It is being very disruptive.

Any ideas / opinions are appreciated.

 

[marclant]

Unfortunately as it fills up the actlog so I can't even read from it I can't get details from the server itself.

Don't confure "active log" with "activity log". The errors you show suggest that the active log fills, causing transactions to fail. This has nothing to do with the activity log. You can't query the activity log because the active log is nearly full and the server runs slow.

I suggest increasing the size of the active log size: https://www.ibm.com/support/knowled...ibm.itsm.srv.doc/t_act_log_size_increase.html