Hello, support We are having a continuing problem in our production environment where attempted ODBC connections return error "Database server not found". Usually, in a few seconds, the connections work again. The DB server itself never goes down. We checked everything that could be checked on the network level, so now we need to know. Are there any known issues with DB server accepting new connections when long update/insert transactions are running? These DBs have ML servers synchronizing data from various locations. Sometimes, the data chunks are very big, it can take up to a minute to commit them. Thank you Arcady |
Does your server run routinely at a non-default port (i.e. not at the default SQL Anywhere port 2638)? Are there other database servers running at the same machine, so another database server uses the default port (and then all further database servers will use the ephemeral ports 49152 and above by default)? Error 10061 is a network (WinSock) error "No connection could be made because the target machine actively refused it". I'd suggest to have a look at that FAQ which deals with many details of connecting, including an explanation on that error. There is more than 1 DB service on this server and there really was no port in the client settings. I set up the port, we shall see if it helps Thank you for now
(18 Jun '20, 04:55)
Arcady Abramov
Replies hidden
In my (limited!) understandig, your client log seems to imply that there is no database server running at the default port 2638. Is that correct? What does "dblocate -dv 10.10.100.64" show? (No need to show that publically!) - Does that fit your expectations?
(18 Jun '20, 05:15)
Volker Barth
|
Have you tried to debug client-wise via -LOG <logfile> how the connection to the database server is found? "Database server not found" does indicate the database server itself is out of reach, IMHO this has nothing to do with a busy database server because it fails before any database connection is established.
OK, and if the error were "Connection timeout" (after, say, 2 seconds), could this be because of DB server being busy? How busy would it have to be to get this result?
Well, I can't tell that but I guess it would help others if you could show a snippet of such a -LOG protocol with failed connections (both when the server was not found or when established connections have been dropped).
Note, you can modify the time out behaviour both on the sercer and client side...
This is from the log. Is it possible to tell if this is a client/network or server error due to service unavailability?