(Sorry for my English, I'm French)
I send my logs via syslog on port 514 or 40514 and I have a problem with this new system.
I send 10 tests (logger) and I receive only 9. If I send 5, I receive 4. If I send 19, I receive 18 (always missing the last one on this new system).
When I use TCPDump, to see what is send and I have always 10 if test is 10, 5 is test is 5..... All my log is send but in the Splunk Search engine, I miss the last? I dont know why, its very strange....
I get more than 50 systems and the problem is only on this new system.
It's the same system for another question here: http://answers.splunk.com/answers/230364/french-syslog.html
(I have 2 UF with VIP, 1 indexer, 1 search engine and 1 for management)
What you thought? Are there other tests that I could do?
Do you have a track for help me to try to correct this problem?
My suspicion is that this is a
buffer flush problem and that the last event will come, but not until something forces the buffer to flush. If you send 5 and only get 4 and then send another 2, do you get the 5th one from the first batch or is it gone? If the former, then there is probably something you can do to force syslog to flush when you need it to (after a certain period of no more input).