\pard\tx960\tx1920\tx2880\tx3840\tx4800\tx5760\tx6720\tx7680\tx8640\tx9600\f2\b0\i0\ul0\fs28 niload fatal error communication failure\
\
Q: In the midst of a rather large
\f0\fs24 niload hosts
\f2\fs28 to the
\b root
\b0 domain, I got the message\
\
\f0\fs24 niload: fatal error: Communication failure\
\
\f2\fs28 I have a large network, with several subnets, and at least two clone servers of the
\b root
\b0 domain on each subnet. What's going on?\
\
A: The “fatal error” looks more frightening than it really is. Chances are the the niload(1) will work correctly if you try it again, though you might have to wait twenty or thirty minutes before the attempt.\
\
Here's what happened.\
\
When you do the niload, NetInfo makes changes to the database. The various clone servers notice that their copies of the database are now out-of-date, and need updating. Each clone ends up requesting a copy of the database from the master; the master gets busy satisfying these requests, and the niload times out.\
\
There are two workarounds for this. You could try the niload again, after waiting a bit. Or, especially if the changes are few, you could use NetInfoManager or niutil to make them.\
\
Note: niload can die in other ways, for example if it's fed a file with syntactically incorrect information. (It won't necessarily die on bad input, though. Bad input may corrupt your NetInfo database to some extent. See the precautions in NextAnswer netinfo.476. Usually it's better to use NetInfoManager and NetManager.)
\i
\i0 If you had bad input, re-do the niload with a -v option and make a fair guess, based on how many +'s get printed, about where the error is. For example, if you see 43 plus signs printed before niload dies, look around line 43 in your data file, using an editor. (Comment lines are not counted.)\
\
QA490\
\
\pard\tx520\tx1060\tx1600\tx2120\tx2660\tx3200\tx3720\tx4260\tx4800\tx5320 Valid for 1.0\
