I have a customer running an accounting package that connects to a MySql data base on a server. 6 workstations connect to the server and run great. Fast and never a problem. However yesterday running the programs on the server directly got so slow that it takes 5 minutes to run a report that takes only 2 seconds on the 6 workstations. It takes 15 seconds to read a single record! This is true for any program that accesses the data base. All other window functions work fine. This happened all of a sudden. Something is wrong with the server but not the workstations (all computers are running XP pro) and I've no idea what is wrong. Any suggestions would be appreciated.
-
-
BTW the're running MySql 4.022 and ODBC 3.51 -
I usually do not give advice to paid consultants, but I'll give it a try.
Ususally the first things you investigate are:
1. changes in the server's software (not necessarily the db server)
2. scheduling algorithm or mechanism for database server degraded (usually by IT dept)
3. database, tables or INDEXES corrupted
4. indexes too large
5. links to clients taken out, so connections must be re-routed
If that is not it, you could try to reorganize the MySQL database, so tables are fixed, indexes are re-built, etc.
That's all I can suggest.
Ronald :cool:Comment
-
Thanks, I don't seem to think the data base itself has a problem since all other workstations are running normal. Also if I use mysql directly all queries work quickly as expected. The software uses ODBC conectivity and I'm wondering if there is some problem between them. I've reinstalled version 3.51 but it did not solve the problem.
Could you elaborate on your option 5?
Originally posted by ronverdonkComment
-
Imagine this network setup.Code:+--------+ +--------+ | node A |<----->| node C |<----------+ +--------+ +--------+ | A | | | V V +--------+ +--------+ +--------+ | node B |<----->| node D |<----->| node X | +--------+ +--------+ +--------+
You usually go from A to B via the A-B link. In case the A-B link goes down, the network people have defined so-called alternate paths. That could mean that, when link A-B is down, you go from A to B via links A-C, C-X, X-D and D-B. Problem here is not the links between the various systems, but the fact that they are configurated as store-and-forward nodes. Meaning that all data coming in is first stored at the node's spool, later scheduled, read from the spool, transferred to the next node and so on.
When there are , say, 10 nodes inbetween your A en B, that is a lot of writing, scheduling, reading and data forwarding.
Hope you see where my comment 5 came from.
However, in your case I think it must be something in the ODBC.
Hope you find it.
Ronald :cool:Comment
-
I have solved this problem!!!! Glory. Some bug somewhere enabled tracing
on the data sources. I saw this comment on some forum. Getting rid of it solved the problem. I've know idea how it got set on.
Thanks all for taking the time to listen and respondComment
Comment