Board index » delphi » IB 6.5 database problem

IB 6.5 database problem


2008-01-14 08:55:07 PM
delphi210
For years we had IB 6.5 server running on an NT4 sp6a primary domain controller. 7 databases, the largest of which was about 600mb. Due to a company buyout, I had to reload the NT4 server as a standalone server. I backed up the databases, and then wiped the server clean and reloaded with NT4, sp6a. Then I updated with all the latest patches. Then I installed IB6.5 and restored the databases.
Since the reload, we have been having issues with the ibserver.exe service spiking the cpu 100%. If clients are connected to the databases, they continue to work (with noticeable slowness), but if new clients try to attach, the applications hangs. All our applications are written in delphi. Some apps use the bde, some up ib components. Nothing major was changed in the applications that we can find that would cause the issues.
After the databases hang, we have to kill the ibserver service and restart it, and then backup and restore the databases. Then clients can reattach.
I am using the same ibconfig file from before the reload and IB6.5 is the only thing this server does.
Any help would be greatly appreciated.
Thanks
Aaron Masser
 
 

Re:IB 6.5 database problem

Aaron Masser writes:
Quote
For years we had IB 6.5 server running on an NT4 sp6a primary domain controller. 7 databases, the largest of which was about 600mb. Due to a company buyout, I had to reload the NT4 server as a standalone server. I backed up the databases, and then wiped the server clean and reloaded with NT4, sp6a. Then I updated with all the latest patches. Then I installed IB6.5 and restored the databases.

Since the reload, we have been having issues with the ibserver.exe service spiking the cpu 100%. If clients are connected to the databases, they continue to work (with noticeable slowness), but if new clients try to attach, the applications hangs. All our applications are written in delphi. Some apps use the bde, some up ib components. Nothing major was changed in the applications that we can find that would cause the issues.

After the databases hang, we have to kill the ibserver service and restart it, and then backup and restore the databases. Then clients can reattach.
I am using the same ibconfig file from before the reload and IB6.5 is the only thing this server does.

Did you reinstate backup routines (running gbak)?
Gbak may have done eanough garbage collection to keep you out of trouble.
--
Aage J.
 

Re:IB 6.5 database problem

Aaron Masser writes:
Quote
After the databases hang, we have to kill the ibserver service and
restart it, and then backup and restore the databases. Then clients
Why must you backup and restore the databases? What error do you get if
you do not backup and restore?
Quote
can reattach. I am using the same ibconfig file from before the
reload and IB6.5 is the only thing this server does.
Have you run gfix on the databases to ensure that they are not corrupt?
What what are the header statistics for a database that is hung?
Are there any errors in the interbase.log file?
--
Bill Todd (TeamB)
 

Re:IB 6.5 database problem

"Aaron Masser" <XXXX@XXXXX.COM>writes
Quote

Is this different hardware? I am not sure about 6.5, but 6 would thrash on a
multi-core/cpu system, so if you re-installed on some more modern hardware
that maybe has multiple cores, this could be part of your slowness problem.
If this is the case, try setting the affinity for ibserver to a single core.
Dave White
 

Re:IB 6.5 database problem

No, this is the same hardware. It is a single processor server. I think I did set the affinity to single core just to rule that out.
"Dave White" <XXXX@XXXXX.COM>writes:
Quote
"Aaron Masser" <XXXX@XXXXX.COM>writes
news:478b69bb$XXXX@XXXXX.COM...
>

Is this different hardware? I am not sure about 6.5, but 6 would thrash on a
multi-core/cpu system, so if you re-installed on some more modern hardware
that maybe has multiple cores, this could be part of your slowness problem.
If this is the case, try setting the affinity for ibserver to a single core.

Dave White


 

Re:IB 6.5 database problem

"Bill Todd" <XXXX@XXXXX.COM>writes:
Quote
Aaron Masser writes:

>After the databases hang, we have to kill the ibserver service and
>restart it, and then backup and restore the databases. Then clients

Why must you backup and restore the databases? What error do you get if
you do not backup and restore?
This is the only way we can get our programs to reattach. If we don't backup and restore the programs just sit at the connection statement in our code.
Quote

>can reattach. I am using the same ibconfig file from before the
>reload and IB6.5 is the only thing this server does.

Have you run gfix on the databases to ensure that they are not corrupt?
I have tried this, however, could you please tell me what switches I should use. Maybe I missed something.
Quote

What what are the header statistics for a database that is hung?
Here are the stastics from this morning..the database was not hung when I took them
Database header page information:
Flags 0
Checksum 12345
Generation 16859
Page size 4096
ODS version 10.1
Oldest transaction 15072
Oldest active 15073
Oldest snapshot 15066
Next transaction 16852
Bumped transaction 1
Sequence number 0
Next attachment ID 0
Implementation ID 16
Shadow count 0
Page buffers 20509
Next header page 0
Database dialect 3
Creation date Jan 14, 2008 9:36:14
Attributes database shutdown
Variable header data:
Sweep interval: 20000
*END*
Quote

Are there any errors in the interbase.log file?
These are the last two errors in the log...
JPANC1DBS2 (Server) Tue Jan 15 07:03:31 2008
INET/inet_error: read errno = 10054
JPANC1DBS2 (Server) Tue Jan 15 08:08:03 2008
INET/inet_error: read errno = 10054
Quote

--
Bill Todd (TeamB)
Thanks
 

Re:IB 6.5 database problem

gfix -validate -full
--
Bill Todd (TeamB)
 

Re:IB 6.5 database problem

tracker.firebirdsql.org/browse/DNET-62
Aaron Masser writes:
Quote

These are the last two errors in the log...

JPANC1DBS2 (Server) Tue Jan 15 07:03:31 2008
INET/inet_error: read errno = 10054

JPANC1DBS2 (Server) Tue Jan 15 08:08:03 2008
INET/inet_error: read errno = 10054
tracker.firebirdsql.org/browse/DNET-62
msdn.microsoft.com/library/default.asp
An existing connection was forcibly closed by the remote host. This
normally results if the peer application on the remote host is suddenly
stopped, the host is rebooted, the host or remote network interface is
disabled, or the remote host uses a hard close (see setsockopt for more
information on the SO_LINGER option on the remote socket). This error
may also result if a connection was broken due to keep-alive activity
detecting a failure while one or more operations are in progress.
Operations that were in progress fail with WSAENETRESET. Subsequent
operations fail with WSAECONNRESET.
Google Results....
www.google.ca/search&meta=
Have you checked to see if your net card has the same driver? Newer,
older etc?
Quote
>--
>Bill Todd (TeamB)


Thanks


--
Will R
PMC Consulting