Subject | Firebird CS Database Corruption |
---|---|
Author | stevemilner_2000 |
Post date | 2005-06-07T11:17:24Z |
If this appears twice, please accept my apologies since I posted this
to Atkin news server and have come to the conclusion that doesn't do
anything.
I was wondering is anyone could help me with this problem, or point me
in the right direction to help myself.
I am using firebird 1.5 Classic server on an SMP (Twin Xeon Processor)
with a Fedora 3 Linux OS as host.
The database is the central data store for retail system for 100
stores with a high product count (80K products) and a high transaction
volume. Many tables have > 8M records. The database is heavily normalised.
We use a threaded kylix3 app (running on a separate apps server) which
is collecting data from the retail stores and populating the central
database on an hourly basis via DSL/ISDN data links.
Although the database doesn't seem to be suffering any problems in
normal use; on a weekly basis gfix -v -full is reporting several
errors with the central database, mainly with index page errors, some
database page errors and the occasional record error. These are
resolved with a gfix -mend and a backup and restore. (Freshly restored
database shows no errors with gfix)
The firebird log contains a few of these error:
(Average 2 per day but often 2 or 3 days without error)
host.domain.net Sun Jun 5 08:19:40 2005
INET/inet_error: read errno = 110
host.domain.net Mon Jun 6 13:28:44 2005
INET/inet_error: read errno = 104
host.domain.net Mon Jun 6 19:28:59 2005
INET/inet_error: send errno = 104
host.domain.net Mon Jun 6 19:28:59 2005
SERVER/process_packet: broken port, server exiting
I have read the tips on avoiding database corruption and have forced
writes set to on, as well as only accessing the database by
host.domain:/path/to/db
Googling for the log errors leads me to beleive these problems are
network related in some way but no further information as to how to
resolve them.
I have the following questions:
Are the errors reported by gfix simply a consequence of the large and
active database and should I schedule a backup and restore every week
to resolve the problems?
Are the gfix errors related to the errors reported in the firebird log
and if so what is the cause of these errors and how do I reduce or
prevent these errors.
Has anyone else suffered from simialr problems and if so what have you
done to resolve this?
I am open to further advice.
If any further information is required please let me know.
Regards,
Steve
to Atkin news server and have come to the conclusion that doesn't do
anything.
I was wondering is anyone could help me with this problem, or point me
in the right direction to help myself.
I am using firebird 1.5 Classic server on an SMP (Twin Xeon Processor)
with a Fedora 3 Linux OS as host.
The database is the central data store for retail system for 100
stores with a high product count (80K products) and a high transaction
volume. Many tables have > 8M records. The database is heavily normalised.
We use a threaded kylix3 app (running on a separate apps server) which
is collecting data from the retail stores and populating the central
database on an hourly basis via DSL/ISDN data links.
Although the database doesn't seem to be suffering any problems in
normal use; on a weekly basis gfix -v -full is reporting several
errors with the central database, mainly with index page errors, some
database page errors and the occasional record error. These are
resolved with a gfix -mend and a backup and restore. (Freshly restored
database shows no errors with gfix)
The firebird log contains a few of these error:
(Average 2 per day but often 2 or 3 days without error)
host.domain.net Sun Jun 5 08:19:40 2005
INET/inet_error: read errno = 110
host.domain.net Mon Jun 6 13:28:44 2005
INET/inet_error: read errno = 104
host.domain.net Mon Jun 6 19:28:59 2005
INET/inet_error: send errno = 104
host.domain.net Mon Jun 6 19:28:59 2005
SERVER/process_packet: broken port, server exiting
I have read the tips on avoiding database corruption and have forced
writes set to on, as well as only accessing the database by
host.domain:/path/to/db
Googling for the log errors leads me to beleive these problems are
network related in some way but no further information as to how to
resolve them.
I have the following questions:
Are the errors reported by gfix simply a consequence of the large and
active database and should I schedule a backup and restore every week
to resolve the problems?
Are the gfix errors related to the errors reported in the firebird log
and if so what is the cause of these errors and how do I reduce or
prevent these errors.
Has anyone else suffered from simialr problems and if so what have you
done to resolve this?
I am open to further advice.
If any further information is required please let me know.
Regards,
Steve