FirebirdSQL
Database stops responding and only backup-restore fix it
Hello,
We have a firebird database and it stops responding every 2-3 days and fails with "Unable to complete network request to host "localhost"".
Firebird restart doesn't fix the problem. The only way to fix the db is to make back-up and then restore it.
We have a .NET application and use EF Core.
Does anybody have ideas how to fix it?
What Firebird version and architecture ? OS ? Did you check firebird.log ? Did you run validation of broken database ? Do you have one for analyze ?
What Firebird version and architecture ? OS ? Did you check firebird.log ? Did you run validation of broken database ? Do you have one for analyze ?
Firebird 2.5 OS - Windows Server 2008
I checked firebird.log. It contains many records like that
INET/inet_error: read errno = 10054, client address = 127.0.0.1/65525, user = АДМИНИСТРАТОР.-1.-1
We didn't run validation.
Here is the database for analyze https://disk.yandex.ru/d/4gMlQoGOqkDaOQ
Thank you!
Sorry, moved to Google Drive - https://drive.google.com/file/d/1rqO9SH-baLjIXt8XBg-MG2-HXqBLE4zF/view?usp=sharing
I see no corruption at your db.
It have stuck OIT and needs to run sweep.
Also, FW=OFF is not recommended.
What exact Firebird version do you use ? And what architecture (SS\CS\SC) ?
When Firebird "stop responding" do you able to make new connection and run simple query, such as SELECT * FROM RDB$DATABASE ?
Run gstat -r and you'll see two tables with a lot of record backversions:
NumberOfVehiclesOnRoute (165)
Primary pointer page: 249, Index root page: 250
Average record length: 28.30, total records: 2557028
Average version length: 10.29, total versions: 415523, max versions: 156
Data pages: 23380, data page slots: 23380, average fill: 67%
OBJECTS (128)
Primary pointer page: 175, Index root page: 176
Average record length: 137.26, total records: 4667
Average version length: 21.65, total versions: 18503156, max versions: 123690
Data pages: 91100, data page slots: 91100, average fill: 96%
Note, the table OBJECTS have record versions ~4000 time more than records !
So, check transactions management in your apps, run sweep regularly and keep an eye on stats. More questions you may ask at our support group https://groups.google.com/g/firebird-support or at https://www.sql.ru/forum/interbase
Thanks for the answer.
I'm not a firebird guru, so could you provide more details?
As for transaction management, all queries are made with EF Core (.NET Firebird client), so not sure how to change it.
As for sweep. Do you recommend to run gfix -s[weep] [-i[gnore]] database_name on nightly basis, e.g. ??
On 2021-10-15 12:13, PavelEmelyanov wrote:
Thanks for the answer.
I'm not a firebird guru, so could you provide more details?
As for transaction management, all queries are made with EF Core (.NET Firebird client), so not sure how to change it.
As for sweep. Do you recommend to run gfix -s[weep] [-i[gnore]] database_name on nightly basis, e.g. ??
The GitHub issues of the Firebird project is intended for reporting bugs, not for support questions.
For support questions on Firebird itself, please use firebird-support (https://groups.google.com/g/firebird-support) as your first port of call. For questions specifically related to using Firebird from .NET (core), use firebird-net-provider (https://groups.google.com/g/firebird-net-provider).
Mark
@PavelEmelyanov please answer all my questions - I'm not asking them for fun. Most likely there are no problem with Firebird itself. To ensure, I need answers on my questions.
To learn about best practices for application development and transaction control, use the sources Mark and I showed above.
I see no corruption at your db. It have stuck OIT and needs to run sweep. Also, FW=OFF is not recommended. What exact Firebird version do you use ? And what architecture (SS\CS\SC) ? When Firebird "stop responding" do you able to make new connection and run simple query, such as
SELECT * FROM RDB$DATABASE?
I'm sorry, I didn't notice your questions. I'll go back with answers