network Connection Timed Out (41)

[JayKoonz]

I have a Windows Master Server (NB 4.5FP6), 2 Windows Media servers that all the Windows clients use for backups, and 5 Solaris media Servers that are only used to backup themselves Lan-Free. One of the Solaris servers constantly times out (41) when it's doing a backup. Doesn't seem to be tied to time-of-day since I can fire it up just about anytime and know it will soon die. The amount of time before it gets a (41) also varies but in all cases, it backup up between 2.92gb and 3.06 gb bfore it gets a "network connection timed out(41)". What network connection is timing out ? Since it's using itself as a media server, who does it have to talk to ? Why always right around 3 gb ? Some buffer filling up somewhere ? It was able to connect okay to start the backup, why time out now ? After the job times out, it will fire itself back up in a while and try again. With the same result, network connection timed out(41) after approx 3 gb.

 

[lenski]

Shaky ground for me but as I remember the media server has to comunicate with the client via the network address. The fact that the client is on the same box makes no differance, the command instructions are the same no matter what the set up. The data path of course is direct but the process to process comunications need to be able to make network conections. Now it may be many reasons why that is failing but I would check all the perameter files. Also check the logs for an indication of the point of failure. Resubmitting of a failed job is normal default but can be changed if you wish. I can't comment on the 2gb of data but I never trust that element of the display.

 

[enterprisebu]

I have several boxes that tend to do the same type of thing. It's very frustrating. We've frequently found that there is a file on a system that hangs up the job. After hearing no response (no data writes) the job decides to time out. Fixing the problem or deleting the file has improved the backups. Then again - there are those sytems we just haven't been able to fix .

 

[mrjazz]

I am not absolutely sure if the following will result in an exit status 41, but:

have you checked already if you have a stale NFS mount on your Solaris machine?

mrjazz

 

[lenski]

We get exactly the same thing but only with corrupt Lotus Notes Databases. To locate the file I set verbose to 5 and then check what the last file to be accesed was before the job crashed.

 

[day88]

lenski, like to check the lotus agent setting you have.

In a Win2k environment and using a Lotus-Notes policy, how to i set the directive to backup notes.ini???
I tried using "NOTES_INI_PATH=c:\lotus\domino\notes.ini" but seem not working. It a non-partitioned domino server.
Thanks

 

[JayKoonz]

What fixed the problem for us was setting "LOCKED_FILE_ACTION = SKIP". Looks like we were hitting alot of locked files in a particular file one right after the other.

 

[lenski]

On the NT side we don't use the Notes Agent, we backup as standard servers. Policy is MS-Windows-NT and file list is ALL_LOCAL_DRIVES. So we don't have to referance the notes.ini file. Beware of the skip files option we had so issues with it last year not working with Notes.

All heavy Notes work is done on Solaris. Which does not help you much But what you put looks okay to me. And it must be fine if it can do several gb before bombing. As I said when ever we have the 41 it's always due corrupt notes files. I turn verbose logging on and check what the last file to be accessed was. I then get the Notes support team to fix the file.

 

[sebasg]

Have you tried changing the CLIENT_CONNECTION and CLIENT_READ TIMEOUTs?
S.-